Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats

Hui Yu; Shilin Zhao; Scott Ness; Huining Kang; Quanhu Sheng; David C Samuels; Olufunmilola Oyebamiji; Ying-yong Zhao; Yan Guo

doi:10.1371/journal.pcbi.1007968

. 2020 Jun 8;16(6):e1007968. doi: 10.1371/journal.pcbi.1007968

Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats

Hui Yu ^1,^*, Shilin Zhao ², Scott Ness ¹, Huining Kang ¹, Quanhu Sheng ², David C Samuels ³, Olufunmilola Oyebamiji ¹, Ying-yong Zhao ⁴, Yan Guo ^1,^*

Editor: Lilia M Iakoucheva⁵

PMCID: PMC7302867 PMID: 32511223

Abstract

Very short tandem repeats bear substantial genetic, evolutional, and pathological significance in genome analyses. Here, we compiled a census of tandem mono-nucleotide/di-nucleotide/tri-nucleotide repeats (MNRs/DNRs/TNRs) in GRCh38, which we term “polytracts” in general. Of the human genome, 144.4 million nucleotides (4.7%) are occupied by polytracts, and 0.47 million single nucleotides are identified as polytract hinges, i.e., break-points of tandem polytracts. Preliminary exploration of the census suggested polytract hinge sites and boundaries of AAC polytracts may bear a higher mapping error rate than other polytract regions. Further, we revealed landscapes of polytract enrichment with respect to nearly a hundred genomic features. We found MNRs, DNRs, and TNRs displayed noticeable difference in terms of locational enrichment for miscellaneous genomic features, especially RNA editing events. Non-canonical and C-to-U RNA-editing events are enriched inside and/or adjacent to MNRs, while all categories of RNA-editing events are under-represented in DNRs. A-to-I RNA-editing events are generally under-represented in polytracts. The selective enrichment of non-canonical RNA-editing events within MNR adjacency provides a negative evidence against their authenticity. To enable similar locational enrichment analyses in relation to polytracts, we developed a software Polytrap which can handle 11 reference genomes. Additionally, we compiled polytracts of four model organisms into a Track Hub which can be integrated into USCS Genome Browser as an official track for convenient visualization of polytracts.

Author summary

Short tandem repeats in the human genome are frequently used as genetic markers in population studies and they are also associated with genetic diseases. Nevertheless, accurate localization and mapping of short tandem repeats in the reference genome have not been well addressed, and there is a lack of systematic catalog of short tandem repeats in the most updated human reference genome. Here, we compiled an updated census of mono-nucleotide/di-nucleotide/tri-nucleotide repeats (MNRs/DNRs/TNRs) from the human reference genome GRCh38, and collectively termed them polytracts. The resultant polytract dataset encompasses TNRs, the presumably more biologically significant repeat species, as well as the under-studied species MNRs and DNRs. With such a composition, the polytract dataset can provide a negative control for genome analyses which are potentially confounded with polytract regions, and it can also be used as a discovery tool to screen for MNR/DNR/TNR-characteristic genomic features. We integrated the polytract dataset with genome coordinates of RNA-editing sites, and found significant enrichment of C-to-U and non-canonical RNA-editing events in adjacency of MNR polytracts, especially break-points of tandem polytracts. The same phenomenon was not observed for the canonical A-to-I RNA editing type. This distinct enrichment patterns between canonical and non-canonical RNA-editing events provides a negative evidence against the authenticity of non-canonical RNA-editing events. Similarly, we examined locations of enhancer sequences relative to polytracts, and found varied degree of locational enrichment among subtypes of polytracts. In practice, different researchers may be interested in different genomic features and/or different organisms, so we developed a tool Polytrap and a Public Track Hub to assist with general locational enrichment analysis of polytracts with respect to localizable genomic features.

Introduction

In the human genome, short tandem repeats (STRs) of unit size 1–6 bp were estimated to occupy up to two million loci [1] or 3% of the genome [2], and they are believed to bear genetic [3], evolutional [4], and pathological [5] significance. Nevertheless, because their repetitive nature inevitably leads to stutter noise in sequencing [6], STRs are often simply treated as suspicious blacklists [7] in practical Next-Generation Sequencing data analyses. An early catalog of human genome STRs dated back to 2003 [1], at which time the human reference genome was GRCh33. Today, the human reference genome has upgraded to GRCh38, five updates in succession to that antique version. In the past decade, although many efforts were dedicated to calling STR variations from Next-Generation Sequencing data, an updated census of STR in the human reference genome is surprisingly unobserved. As a preparation step to their major analyses, two studies [8, 9] compiled STR catalogs in GRCh37 and GRCh38, respectively; both works employed ad hoc procedures and custom parameters, thus generating highly customized datasets unsuitable for general use. In particular, both catalogs were initially generated by an inference-based algorithm TRF [10], so by definition they represented a mixture of perfect and imperfect STRs.

Among all STR species, trinucleotide repeats (TNRs) are the ones most likely to be implicated in human genetic diseases [5], and thus TNRs have unsurprisingly attracted more research efforts [11, 12] than other STR species. Much fewer research efforts have been dedicated to analysis of the shortest repeat species–mono-dinucleotide repeats (MNRs) and di-nucleotide repeats (DNRs). However, a recent discovery that drosophila enhancers are characterized by DNRs [13] hints at the biological significance of very short tandem repeats. Additionally, we found that, in the human mitochondrial genome, MNRs or DNRs tend to over-represent tri-allelic heteroplasmy and RNA-DNA differences (RDDs) [14]. The presence of abundant RDDs may result from RNA-editing, a post-transcriptional regulatory mechanism, but can also represent technical artifacts in both DNA and RNA sequencing. As reviewed earlier [15], RDDs observed in RNA-Seq data comprise real RNA-editing events, single-nucleotide polymorphisms, and artifacts (errors). Actually, there is a strong voice that non-canonical RNA editing events are primarily attributed to sequencing errors or bioinformatics pitfalls [16]. Very short tandem repeats, such as MNRs and DNRs, pose higher sequencing/analysis challenges than other genomic regions, so RDDs around very short tandem repeats may be especially suspicious.

Intrigued by the dispute over non-canonical RDDs [15] and their involvement in mitochondrial MNRs and DNRs [14], we felt a pressing need to expand the analysis of repeat-associated genomic features to the whole contemporary human reference genome (GRCh38), aligning the under-investigated MNRs and DNRs with the more established TNRs. To this end, we first created an up-to-date catalog of precisely defined very short tandem repeats, including MNRs, DNRs, and TNRs, which are termed “polytracts” in general. Different from related works, our polytract catalog was intentionally biased towards very short tandem repeats which comprise MNRs, DNRs, and TNRs, and we included only perfect matches to the polytract definition (see Materials and Methods). After creating the polytract catalog, we revisited prior DNR/TNR-related findings using our updated data. Finally, we performed locational enrichment tests of polytract regions against nearly a hundred genomic features, with an emphasis on RDDs. The present work led to an updated census of MNRs/DNRs/TNRs in human reference genome GRCh38, corroborated and expanded biological significances associated with DNRs and TNRs, and revealed landscapes of genomic features enriched within MNRs/DNRs/TNRs. A software program named Polytrap was developed to assist with locational enrichment analysis of polytracts with respect to localizable genomic features in the human genome (GRCh37 and GRCh38) and other model organisms.

Results

A census of MNR/DNR/TNR polytracts in human genome

As detailed in Materials and Methods, we identified MNRs/DNRs/TNRs spanning at least six (for MNR) or three (for DNR and TNR) units in the human reference genome GRCh38, and obtained their summary statistics at genome level (Table 1). MNRs, DNRs, and TNRs occupy 59.1 million (1.9%), 70.6 million (2.3%), and 14.7 million (0.5%) nucleotides, altogether consuming 144.4 million nucleotides (4.7%) of the human reference genome GRCh38 (Fig 1A). Simply comparing the numbers, this latest percentage of 4.7% is much larger than previous estimations of ~3% that were based on early reference genomes [1, 2]. Because the minimum length of STRs was not set identically and the encompassed STR species were not the same, these percentages were not directly comparable to each other. However, we were able to compare these statistics between GRCh38 and GRCh37 in parallel. Using the same protocol, we identified all polytracts in the reference genome GRCh37. Surprisingly, while the total genome size reduced by 0.2% from GRCh37 to GRCh38 [17], the nucleotide volume occupied by MNRs, DNRs, and TNRs consistently increased in absolute number (S1 Table); in total, GRCh37 and GRCh38 harbor 141.6 million (4.6%) and 144.4 million (4.7%) polytract nucleotides, respectively. It seemed that broader regions of polytracts are revealed in newer, more refined reference genomes.

Table 1. Summary statistics of three clades of polytracts in GRCh38.

Polytract species	Number of polytracts	Nucleotides enclosed in polytracts	Percentage of polytract nucleotides in genome	Mean length	Median length
A/T	7,119,220	55,290,931	1.79%	7.8	6
C/G	610,474	3,839,875	0.12%	6.3	6
MNR total	7,729,694	59,130,806	1.91%	7.6	6
TA	2,764,278	19,948,282	0.65%	7.2	6
CT/GA	3,679,922	25,030,351	0.81%	6.8	6
CA/GT	3,371,036	25,125,048	0.81%	7.5	6
GC	71,160	476,554	0.02%	6.7	6
DNR total	9,886,396	70,580,235	2.29%	7.1	6
AAT	353,551	3,828,512	0.12%	10.8	9
ACC	238,068	2,302,369	0.07%	9.7	9
AAG	183,579	1,859,914	0.06%	10.1	9
AGG	183,232	1,856,718	0.06%	10.1	9
AAC	146,444	1,720,479	0.06%	11.7	10
CAG	131,235	1,290,213	0.04%	9.8	9
ATC	123,051	1,223,622	0.04%	9.9	9
ACT	36,083	352,774	0.01%	9.8	9
CGG	21,508	238,039	0.01%	11.1	10
GAC	1,396	13,915	0.00%	10.0	9
TNR total	1,418,147	14,686,555	0.48%	10.4	9

Open in a new tab

In the field, there is a similar data resource of GRCh38 DNRs/TNRs available as UCSC Microsatellites (http://genome.ucsc.edu/cgi-bin/hgTrackUi?hgsid=104818371&c=chrX&g=microsat). We extracted a comparable subset of DNRs and TNRs from our polytract dataset and compared it with the UCSC Microsatellite dataset. We found that literally 100% of UCSC Microsatellites were recovered in the polytract set, but the latter contained additional qualified repeat instances missed by the former (S2 Table).

There is evident disparity in the length of different polytract species (Fig 1B). The mean length of a MNR is 7.6 nt, however the two species of A/T and C/G form two distinct groups (mean 7.8 nt vs. 6.3 nt). The mean length of a DNR is 7.1 nt, but two length groups are visually separable: TA and CA/GT species have longer lengths (mean 7.2~7.5 nt), while CT/GA and GC species have shorter ones (mean 6.7~6.8 nt). The mean length of a TNR is 10.4 nt, with three species showing appreciably longer mean length (AAC 11.7 nt, CGG 11.1 nt, and AAT 10.8 nt).

Even more evident disparity is found with tract density, defined as number of polytracts per Mb (Fig 1C). For MNRs, A/T polytracts are 11 times as dense as C/G polytracts. For DNRs, the densest species is the TA polytract, which is interspersed 41 times as dense as the sparsest species, GC. For TNRs, AAT (density 105/Mb) is the densest species, whereas GAC (0.52/Mb) and CGG (12/Mb) are the sparsest species. A statistical significant linear correlation between mean length and density across chromosomes was found only for TNR tracts but not for MNR or DNR (S1 Fig).

At times, two tandem stretches of polytracts are separated by only one nucleotide, and we term these single-site breakpoints “hinges.” There are in total 472,716 hinge sites in GRCh38. As expected, hinge-connected duplexes demonstrate largely the same prevalence order as the polytract unit species per se (Fig 1D), with A/T-joining hinges being the most frequent and C/G-joining hinges the rarest.

Based on their location, polytracts were assigned to seven different genomic region categories, and the distribution of each polytract species across the seven region categories was obtained (S2 Fig). Overall, 90.5% of polytracts are located in intronic (36.8%) or intergenic (53.7%) regions, and this quantity does not vary much among the three polytract clades (MNR 90.9%, DNR 90.3%, and TNR 88.1%, respectively; Fig 1E). Notably, TNR species including GAC, CGG, and CAG, as well as the GC DNR, have substantially elevated exonic portion compared to the general pattern (Fig 1F). The same three TNR species were found most over-represented in exome in an early survey [11]. Of note, GAC and CGG are the two least frequent TNR species in the genome, having chromosome density of 0.52/Mb and 12/Mb, respectively (Fig 1C). The CGG TNR and the GC DNR (Fig 1F) also have notable over-representation in up/down-stream segments (1 kb proximity to gene body) and 5’-UTR regions, a phenomenon rarely seen in other polytract species (S2 Fig).

Corroborated and extended biological relevance of DNR and TNR

Previously, a study of thousands of enhancer sequences in three Drosophila cell lines proposed that DNRs are a general enhancer feature, especially with respect to universal enhancers [13]. Herein, of all 5,967 human enhancers, we found 4,677 (78.4%) embedded or adjoined at least one polytract. Each individual clade, MNR, DNR, and TNR, as well as the whole polytract set, were significantly embedded in enhancers (Bonferroni-adjusted p < 0.01, hypergeometric test; Fig 2A). As shown in Fig 2B, of MNRs, C/G polytracts were more likely to be embedded in enhancers than A/T polytracts (Relative Risk (RR): 84.6 vs. 11.0); of DNRs, GC polytracts were most likely to be embedded in enhancers with a striking RR of 173.6; of TNRs, CGG, AGG, and CAG polytracts stood out ahead of other species, showing RRs 173.0, 85.4, and 68.9, respectively. The TA DNR showed the weakest trend of being embedded in enhancers among the four DNR species (RR = 4.9), which is accordant with the observation in Drosophila [13].

Compared with MNRs, DNRs were more likely to be embedded in enhancer sequences (RR 22.8 vs. 16.0, Fig 2A), and the embedding tendency in DNR was stronger for universal enhancers than specific enhancers (Fig 2C). These observations were again coherent with findings in the Drosophila study [13]. Moreover, we noted an even stronger tendency for enhancers to embed TNRs (RRs 30.9, Fig 2B). To our best knowledge, this is the first time that TNR is identified as a feature of enhancer sequences, especially in the human genome.

Previously, a systematic genomic investigation [11] was conducted on TNRs in the human genome (GRCh36), where the minimum repeat length was set at 18 nt (or “6U” for 6 repeat units). To align our data with the precedent design [11], from the total 1,418,147 TNR tracts, we identified a subset of 39,937 long polytracts that each contained six or more (complete or incomplete) trimer units (i.e., length ≥ 18nt). By repeating the same precedent analytics [11], we revealed largely the same length distributions for the ten TNR species, and rendered a similar TNR clustering based on a matrix of pairwise Kolmogorov-Smirnov test statistics (S3 Fig).

In addition to long repeat tracts of 6 units or more [11], our full TNR polytract set contains a dominant portion (97.2%) of five-unit long or less, which had not been touched on yet. We repeated the same analysis on this tremendously larger, full TNR set, discriminating difference in tract length distributions among TNR species. Interestingly, this moderate-repeat-dominated dataset discriminated two TNR types inconspicuous in the prior study, namely CGG and AAC (Fig 3A and 3B). It is visually discernable that these two TNR species have the highest portions of long repeats (6U or more) than do all other species (Fig 3A). AAC has the most dissimilar length distribution, with the greatest portion (61.8%) of tracts enclosing ten or more nucleotides; no wonder it has the longest average length (Fig 1B). Another peculiar phenomenon with AAC is that the polytract frequency does not decrease monotonously with increasing polytract length; two surprising modes appear at 11 nt and 14 nt, the lengths awaiting a single nucleotide to form intact TNR cycles. Investigating further, we found pre-cycle length modes repetitively appear in a majority part (3~17 U) of the whole length range (3~19 U) of AAC tracts (Fig 3C).

Fig 3 — A, polytract length distribution dissected by TNR species. B, heatmap graph showing the Kolmogorov-Smirnov statistics D resulting from pairwise comparisons of polytract lengths among all 10 TNR subtypes. Hierarchical dendrogram highlights CGG and AAC as unique from the rest majority. Distance metric used one minus correlation value and linkage choice was the average method. C, full-range length distribution of AAC polytracts. Dark black bars highlight periodical pre-cycle modes at length 11 bp, 14 bp, 17 bp, etc..

Non-canonical RNA-editing events are enriched within MNR adjacency or hinge sites

We obtained genome locations for 4,688,495 RNA editing events, which were categorized to three classes: A-to-I, C-to-U, and non-canonical. Because in Next-Generation Sequencing experiments inosine (I) and uridine (U) are replaced with guanine (G) and thymine (T), respectively, we renamed the two canonical classes as A-to-G and C-to-T, respectively (Table 2). Through the binomial probability model (details in Materials and Methods), we examined the locational enrichment tendency of various RNA-editing classes in various polytract clades or species. In a nutshell, we evaluated if RNA editing events were over-represented in the polytract territory as compared to the baseline frequency across the whole genome. Here, the polytract territory referred to the aggregate segments of the genome occupied by all polytracts, and it was their collective nucleotide volume that we quantified. Notably, in addition to the exact territory of polytracts, we extended each polytract one nucleotide bi-directionally and thus formed an adjacency-extended polytract territory. Locational enrichment analysis was performed in parallel between the exact polytract territory and the extended polytract territory.

Table 2. RNA editing event classes.

Practical class name	Standard class name	Nucleotide changes	Incidence
A-to-G	A-to-I	A>G, T>C	4,677,846
C-to-T	C-to-U	C>T, G>A	5,006
non-canonical	non-canonical	All changes except A-to-G and C-to-T	5,643

Open in a new tab

Of all three polytract clades, MNR displayed the most disparate enrichment pattern across RNA editing classes. In terms of the exact polytract territory, MNR enriched non-canonical editing events (Binomial p<1e-22) only, but in terms of the extended polytract territory, MNR enriched both C-to-T events and non-canonical events (Fig 4A, “MNR” panel). All classes of RNA-editing events, including A-to-G, C-to-T, and non-canonical, showed elevated RR of residing in hinge sites (Binomial test p<1e-22 for all; Fig 4A, “Hinge” panel), and the RRs with C-to-T (RR = 380) and non-canonical (RR = 225) were much higher than A-to-G (RR = 3.6). Tandem duplexes of A/T MNRs accounted for a predominant portion of these RNA-editing–concurrent hinge sites—90.4% for A-to-G, 99.3% for C-to-T, and 96.2% for non-canonical. It might be that an actual longer, intact A/T fragment is incorrectly broken into two pieces in the reference genome, which leads to spurious reports of C-to-T RDDs in hinges of A/T MNRs.

Fig 4 — A, A-to-G and C-to-T editing events do not have a higher likelihood of appearing in MNRs, whereas non-canonical events do; A-to-G, C-to-T, and non-canonical RNA-editing events all have a higher likelihood of appearing in polytract hinges. *Exact*, exact regions of polytract occupancy. *Extended*, polytracts extended with one neighboring nucleotide bi-directionally. *Hinge*, single-nucleotide sites located between two neighboring stretches of exact polytracts. Asterisk (*), binomial enrichment p<0.01. B, heatmap of Relative Risk (RR) of polytract bearing an RNA-editing event, depicted for combinations between editing event classes and polytract species. Non-elevated relative risks (RR<1) or statistically insignificant (p>0.01) associations of editing events were imputed with RR = 1.

Considering the nucleotide substitution source and target in C-to-T editing events, we may regard C/G MNRs as the “source” polytracts and A/T MNRs as the “target” tracts. Therefore, it is expected to see C-to-T events condense in the source, C/G MNR (Fig 4B,”exact tract” panel). With bi-directional 1-nt extension, however, C-to-T events demonstrate significant over-representation in both the source, C/G, and the target, A/T (Fig 4B, “adjacency-extended tract” panel). The spurious enrichment of C-to-T events in adjacency of A/T polytracts may also be explained if we presume adjacency of A/T polytracts bears more mapping errors than the reference genome average. While the presumption of polytract hinges and adjacencies bearing more mapping errors is intuitively plausible, wet-lab experiments are needed to verify these suppositions and possibly to discriminate false C-to-T editing events.

Overall, A-to-G events did not show evident locational preference of any tract species, except for a weak enrichment tendency within polytract hinge sites (RR = 3.6). None of the three classes of RNA-editing events was found enriched within DNRs (Fig 4A and 4B). Occasional locational enrichment took place between certain editing classes and TNR species (Fig 4B). Notably, the AAC polytract, a moderately prevalent TNR species with distinctive pre-cycle length modes, was significantly enriched with non-canonical editing events only (RR = 7.1, binomial test p = 4.2e-5, Fig 4B). We speculate that many non-fully-cycled AAC tracts have wrong trailing single nucleotides, which led to the counter-intuition pre-cycle length modes (Fig 3C) and clustering of spurious non-canonical RNA-editing events (Fig 4B). In future refinement of human reference genome, special attention should be given to suspicious trailing sites to incomplete AAC TNRs.

Landscape of genomic feature locational enrichment within polytracts

Moving beyond enhancers and RNA-editing sites, we interrogated polytract locational enrichment with respect to miscellaneous genomic features, including categorized gene regions, RNA-binding protein’s presumable binding segments [18], polymorphism variants [19], somatic mutations [20], etc., thereby resulting in landscapes of polytract locational enrichment against 94 individual genomic features (Fig 5, S3 Table). Globally, TNRs are associated with the largest number of genomic features, and DNRs the least. MNR-enriched features include non-canonical RNA-editing events, enhancers, non-coding RNAs, polymorphism variants, eQTLs, histone modification sites, retrotransposons, and binding segments of several RNA-binding proteins. Different from MNRs, TNRs show over-representation for certain genes (protein_coding genes, pseudogenes, and immunoglobulin genes), but not for any kind of RNA-editing events. DNRs are not associated with either RNA-editing events or gene regions; a specific small RNA (“sRNA”) stands out only because of the dimer-rich paralogs of Clostridiales-1 RNA (RF01699).

Fig 5 — Features manifested in red are deemed statistically significantly enriched. Because nearly 100 features were tested simulatenously, we applied bonferroni correction and deemed p<1e-4 as statistically significant.

We also noted several interesting phenomena in relation to specific genomic features. Exome analysis blacklist variants are enriched within all three clades of polytracts, and the statistical significance indicates such an order: MNR (p = 2.2e-217) > DNR (p = 1.6e-106) > TNR (p = 7.3e-28). Transcription factors (“TFBS”) have a tendency to bind into MNR, DNR, and TNR polytracts, whereas most RNA-binding proteins do not, except for selected TNR-favoring ones, especially splicing factors. Polymorphism variants, including single nucleotide variation (“gnomAD_SNV”) and indels (“gnomAD_indel”), tend to cluster around any clade of polytracts, whereas cancer somatic mutations are only over-represented in MNRs and TNRs. Within somatic mutations, whereas indels are over-represented in both MNRs and TNRs, SNVs (single nucleotide variations) are enriched in TNRs only. Because our somatic mutation data were severely biased towards exome and certain TNR species were over-represented in exome, we were afraid the exclusive enrichment of somatic SNVs in TNR was merely a consequence of these two known biases. So we identified the subset of non-coding somatic SNVs from the total somatic SNVs and performed locational enrichment analysis on them only. Still, non-coding somatic SNVs displayed locational enrichment in TNRs only (S4 Fig).

A software package to evaluate genomic feature enrichment within polytracts

To assist with general locational enrichment analysis with respect to any custom genomic features, we developed a Linux package named Polytrap. With input genomic intervals specified as start and end coordinates on chromosomes, Polytrap distinguishes intervals overlapping with any instance of polytracts, and assesses statistical significance of interval enrichment in polytract clades and species (Fig 6). Our bulk locational enrichment analysis against 94 individual genomic features, as expounded above, was assisted by Polytrap. We recorded the time and memory usage pertinent to RNA-editing analyses (Table 3) to enable estimation of practical computational costs, which is proportional to the size of the input dataset.

Fig 6 — Polytrap stores chromosome locations of polytracts (1-mer, 2-mer, and 3-mer tandem repeats, termed MNR, DNR, and TNR) in a reference genome, and calculates the probability of enrichment of user-given genomic feature within polytracts. For a genome of totally N nucleotides and a polytract territory of n nucleotides, the probability of k out of M instances of a genomic feature being located within tracts is calculated through the binomial model. When the overlapping mode is switched from “singleton” to “multiplex,” k and M designate number of aggregate nucleotides instead of instances (bottom left part).

Table 3. Polytrap computational cost on example sessions.

Sessions were tested on a Linux Ubuntu work station with Intel Xeon CUP E5-2650 V4 @ 2.20GHz and 32 GB memory.

Analysis modality	Input dataset size^*	Time usage	Memory usage
TNR	5K	0m16s	1.7G
TNR	5M	1m14s	1.7G
DNR+TNR	5K	1m43s	7.6G
DNR+TNR	5M	2m49s	7.6G
MNR+DNR+TNR	5K	3m24s	14.2G
MNR+DNR+TNR	5M	4m35s	14.2G

Open in a new tab

*dataset size refers to the number of nucleotide positions involved in the input data file. In these example sessions, 5K corresponds to our C-to-T dataset (5,005 positions) and 5M corresponds to our A-to-G dataset (4.7 million positions).

While some existent genomic analysis tools bear partial capabilities in relevance to STR, Polytrap appears as the only tool to identify and assess the locational enrichment of user-defined genomic intervals within very short tandem repeats in a systematical and straightforward manner (A theoretical comparison is provided in S4 Table). Polytrap allows to focus on subsets of polytracts with or without adjacency extension, or polytracts embedded by particular gene body regions (protein-coding genes, lncRNAs, or pseudogenes). While all analyses presented above were testing over-representation of a genomic feature, we ensured Polytrap have the flexibility of testing under-representation as well. Taking advantage of this option, we investigated the under-representation tendency of different classes of RNA-editing events in polytracts, and found A-to-G events were under-represented in almost all polytract species, excluding A/T, ACC, and hinges (S5 Fig).

Beyond the human genome, STRs receive increasing attention in non-human organisms as well, including macaques [21], dogs [22], fugu [23], and plants [24]. To maximize the potential of Polytrap, we made it capable of handling nine organisms (human, macaque, mouse, rat, dog, chicken, zebrafish, fruitfly, and yeast) and allowed it extendable to uncovered organisms with provisions of Bioconductor’s genome support (DOI: 10.18129/B9.bioc.BSgenome). Polytract data files for four organisms (human, mouse, rat, and fruitfly) are formatted into a Track Data Hub [25] for convenient visualization at UCSC Genome Browser (http://genome.ucsc.edu/). The software package and related data files are released on GitHub (https://github.com/hui-sheen/polytrap/) and our project webpage (http://innovebioinfo.com/Annotation/Polytracts/Polytract.html).

Discussion

STRs are receiving increasing appreciation for their roles as genetic markers, polymorphism variants, origins of heritable neurological diseases [3]. Recently, several analysis tools were developed to facilitate STR variant calling from sequencing data [6, 26, 27]. Among the increasing research efforts, we noted a lack of systematic census and characterization of STRs in the updated human reference genome. A careful scrutiny and characterization of STRs in the reference genome will benefit the expanding STR-focused personal genome analyses. In this work, we found 4.7% of the human genome (GRCh38) is occupied by polytracts, and there are 0.47 million single nucleotides located between tandem polytracts, which we call polytract hinges. More than a half of these hinge sites break apart thymine-involved polytracts. Unlike other TNR species, the AAC polytract shows suspicious pre-cycle modes in length distribution. It is a commonsense that STR boundaries bear a higher mapping error rate than the genome baseline, and our results might further pinpoint the hinge sites and AAC polytract boundaries may bear an exceptionally higher mapping error rate.

In accordance with its inherent error-prone nature, polytracts demonstrate enrichment of non-canonical RNA-editing events, somatic indels, and blacklist variants. Thus, polytracts can reasonably serve as a negative reference set for quality controlling genome/exome sequencing studies, helping to flag suspicious variant calls or other sequencing results that are significantly biased to polytract regions. For this negative quality control purpose, we suggest more weights be given to MNR than DNR and TNR, because monomer runs are especially intractable regions in sequencing experiments thus a lower data quality may be associated with MNR in terms of both the mapped sequence in the reference genome and the re-mapped sequence in a personal genome. Across the comparative enrichment landscapes of MNR/DNR/TNR, somatic SNVs are enriched in TNR only whereas somatic indels are enriched in all three polytract clades. In practical sequencing projects, downstream analyses in succession to variant calling usually trust SNVs more than indels, and a component of too many indels raises a warning sign. The disparate polytract enrichment pattern between SNV and indel is consistent with our varied confidence towards SNV and indel in sequencing practices. Likewise, the blacklist variants display the highest statistical significance of enrichment in MNR yet the lowest statistical significance in TNR, again discriminating the more error-prone nature of MNR. So, it is advisable that sequencing practitioners include polytract/MNR enrichment analysis in their quality control protocols.

Polytracts comprise DNRs and TNRs in addition to MNRs. TNRs are appreciated more as biologically significant motifs than avoidable pitfalls. In the polytract enrichment landscapes, TNR does show significant enrichment for many meaningful cis-elements, including transcription-factor binding sites and RNA-binding protein target segments. Interestingly, many genomic features are enriched in MNR/DNR as well as in TNR. Especially, DNR resembles TNR in demonstrating strong over-representation of enhancers and strong under-representation of A-to-I RNA-editing events, thus indicative of remarkable biological significance. MNR also shows significant enrichment for some meaningful genomic features, including binding targets for select RNA-binding proteins. RNA-binding proteins HNRNPC, TIA1, and U2AF2 all favor U-rich binding motifs and their target sequences show distinctively lower entropy than common RNA-binding proteins [28]; concordantly, binding segments of these three RNA-binding proteins enrich MNR yet not DNR or TNR (Fig 5). In this context, the enrichment within MNR indicates differential characteristics of binding motifs of select RNA-binding proteins. Therefore, it risks neglecting genuine biological signals if we always ignore or downweight features/entities that are enriched in simple repeats such as MNRs. Generally speaking, domain knowledge and meticulous inspection are helpful for interpreting polytract enrichment results rationally.

RNA-editing events are classified as canonical events and non-canonical events, and A-to-I events dominate over C-to-U events in the canonical category. There has been a long-lasting dispute over the authenticity of non-canonical editing events [15]. Peer researchers pointed out flaws in designs or protocols of studies that advocate the universality of non-canonical editing events [29, 30]. Here, we discovered that non-canonical RNA-DNA differences are overrepresented in MNR adjacent sites or hinge sites whereas A-to-I events are generally under-represented in polytracts. From a novel angle, the selective enrichment of non-canonical RNA-editing events within MNR adjacency provides a negative evidence against their authenticity, and it also supports the practical operation of assessing the quality of an RDD dataset by the proportion of A-to-I events [31]. We even argue that a considerable portion of C-to-T editing events are false discoveries, as C-to-T editing events are condensed in the vicinity and hinge sites of polytracts, resembling non-canonical events but not A-to-I events. Our suspicion on C-to-T editing events might be linked to a recent finding that human C-to-U coding RNA editing is largely nonadaptive and that they probably manifest cellular errors [32]. The predominant canonical editing class, A-to-I, is under-represented in all polytract species except A/T MNR, ACC TNR, and hinges. Again, it is alerted that special caution should be cast on polytract hinge sites and boundary regions of certain repeat types.

STRs constitute an abundant component of genomic DNA in not only humans but also many other species, such as macaques [21], fugu [23], and plants [24]. Our work developed a convenient tool, Polytrap, to aid with locational enrichment analysis with respect to polytracts in the reference genomes of diverse model organisms. Polytrap can be used as a quality control tool for genome analyses which are potentially confounded with polytract regions, or as a discovery tool to screen for particular MNR/DNR/TNR polytracts characterized with the user-interested genomic features.

Conclusions

Our work resulted in a comprehensive census of MNR, DNR, and TNR in the latest human reference genome (GRCh38) and a software Polytrap to assist with general locational enrichment analysis of polytracts with respect to localizable genomic features. We found a plethora of genomic features are significantly co-localized with polytracts, including both meaningful biological signals (enhancers, transcription factor binding sites, etc.) and artifact-beset entities (insertions/deletions, blacklist variants, etc.). Most notably, non-canonical RNA editing events are enriched inside and/or adjacent to MNRs, whereas A-to-I editing events are generally under-represented in polytracts. This provides a negative evidence against the authenticity of non-canonical RNA editing events. Polytracts can be referenced as a negative quality control for sequencing studies, but they also bear the potential to inform genomic researches. Polytrap enables swift locational enrichment analysis of subtyped polytracts, and we encourage knowledgeable and meticulous inspection in interpreting the polytract enrichment results.

Materials and methods

Identification and categorization of polytracts

In a reference genome, a polytract is defined as a tract of mono-nucleotide, di-nucleotide, or tri-nucleotide tandemly repeated motifs, with possibly incomplete terminal motif included, where the minimum number of repeated units are 6 for MNR and 3 for DNR and TNR. Unlike many related works which allow fuzzy STRs, we seek only perfect matches to our definition. This was done through a string matching between a pattern and each chromosome sequence, with assistance from R packages “stringr” (https://CRAN.R-project.org/package=stringr) and “BSgenome.Hsapiens.UCSC.hg38” (www.bioconductor.org; DOI: 10.18129/B9.bioc.BSgenome). Because of the complementarity between a purine and pyrimidine pair (A:T and G:C), we combined poly-A and poly-T runs into an “A/T” group, poly-G and poly-C runs into a “G/C” group, poly-CT and poly-GA runs into a “CT/GA” group, and poly-CA and poly-GT runs into a “CA/GT” group. Similarly, a nominal full set of 60 TNR motifs were merged into 10 groups: AAC (accommodating AAC, ACA, CAA, GTT, TGT, and TTG), AAG (AAG, AGA, GAA, CTT, TCT, and TTC), AAT (AAT, ATA, TAA, ATT, TAT, TTA), ACC (ACC, CCA, CAC, GGT, TGG, GTG), GAC (GAC, ACG, CGA, GTC, CGT, and TCG), ACT (ACT, CTA, TAC, AGT, TAG, and GTA), CAG (CAG, AGC, GCA, CTG, GCT, and TGC), AGG (AGG, GGA, GAG, CCT, TCC, and CTC), ATC (ATC, TCA, CAT, GAT, TGA, and ATG), and CGG (CGG, GGC, GCG, CCG, GCC, and CGC). In summary, our resultant polytract dataset consisted of two types of MNRs, four types of DNRs, and ten types of TNRs (Table 1). All instances of GRCh38 polytracts are stored as data files in our public software Polytrap (https://github.com/hui-sheen/polytrap/ and http://innovebioinfo.com/Annotation/Polytracts/Polytract.html).

In the same manner as we dealt with the reference genome GRCh38, we identified all polytracts in the reference genome GRCh37. In our miscellaneous analyses, certain data resources (including enhancers and RNA editing sites) were based on GRCh37 instead of GRCh38, and in such cases we performed the analyses against GRCh37 polytracts.

Locational enrichment analysis of polytracts with respect to localizable genomic features

Having located each polytract in the reference genome, we sought to evaluate the degree of enrichment of a particular genomic feature within the polytract territory, where the genomic feature is represented as a set of genomic sites or intervals and the polytract territory refers to the aggregate segments of the genome occupied by all polytracts (Fig 6). This is a general question of assessing the (unexpected) locational overlapping between one set of genomic intervals and another set of genomic intervals, and previous studies [33–35] have provided solutions of a largely same principle. Basically, we define the expected frequency of observing the genomic feature within the polytract territory as n/N, where N denotes the total size of the genome and n the size of the polytract territory; when k out of M instances of the genomic feature fall into the polytract territory, if the observed frequency k/M is much higher than n/N, it indicates probable enrichment. Classical probability models such as hypergeometric model and binomial model are frequently used in such scenarios to test for locational enrichment. Here, we followed early examples [33, 35] to employ the binomial model.

H0 (null hypothesis)—the concerned genomic feature is distributed in polytract territory at a frequency no higher than random expectation.
H1 (alternative hypothesis)—the concerned genomic feature is distributed in polytract territory at a frequency higher than random expectation.

Central to the hypothesis test was identification and quantification of overlaps between polytracts and instances of the genomic feature. We conceived of two modes for this issue. Under the default “singleton” mode, we regarded a genomic interval as a “singleton” unit, so whenever a spatial overlap appeared between a tract and an instance of the genomic feature, we counted it as ONE overlap. Under the alternative “multiplex” mode, we regarded a genomic interval as a union of its constituent nucleotides, so, when a spatial overlap appeared, we counted MULTIPLE nucleotides towards the overlap quantity (Fig 6).

Under the null hypothesis and the singleton mode, for a genome of totally N nucleotides and a tract territory of n nucleotides, the probability of k out of M instances of a genomic feature being located within tracts was calculated through the binomial model (Eq 1). The Relative Risk of polytract bearing the concerned genomic feature was obtained by dividing the tract-ridden rate by the tract occupancy rate (Eq 2).

p = \sum_{i = k}^{n} (\binom{n}{i}) {(\frac{M}{N})}^{i} {(1 - \frac{M}{N})}^{n - i}

Eq 1

R R = \frac{k / M}{n / N}

Eq 2

When we interrogated polytract association with various types of gene regions and RNA-binding protein target segments, we imposed the multiplex overlap mode. There, the probability of null hypothesis was calculated using Eq 1 too, however in such cases M referred to the total nucleotide quantity of the genomic feature and k denoted the quantity of overlaid nucleotides in all overlapping instances.

By default, Polytrap conducts a one-tailed test through the binomial distribution model, testing for the over-representation tendency of a genomic feature within the polytract territory (Eq 1). However, we ensured Polytrap have the flexibility of testing for the under-representation tendency as well. By setting the directionality option to under-representation, Polytrap calculates the probability of reaching no more than the observed locational overlapping instances (Eq 3), thus effectively assesses the under-representation tendency of a genomic feature within the polytract territory.

p = \sum_{i = 0}^{k} (\binom{n}{i}) {(\frac{M}{N})}^{i} {(1 - \frac{M}{N})}^{n - i}

Eq 3

Lastly, it is worth noting that, in assessment of feature enrichment within polytract territory, we extended each polytract instance to the single immediate neighboring nucleotide bi-directionally. This operation was inherited from our previous related study [14] and it was set as a default but suppressible choice in Polytrap.

Genomic features enrolled in the locational enrichment analysis

Having identified the polytracts in GRCh38 and GRCh37, respectively, we went on to investigate possible locational enrichment tendency of polytracts with respect to an array of genomic features. Ninety-four genomic features localized in GRCh38 or GRCh37 were coarsely grouped to five major classes, namely RDD, enhancer, Gene region in HG38, RNA-binding protein’s binding segments, and miscellaneous (S3 Table). We recruited RDD foremost because RDDs displayed an intriguing enrichment in mitochondrial polytracts [14] and we wanted to verify if the same phenomenon exists in the nuclear genome. Enhancer was chosen because an early study discovered characterization of drosophila enhancer with DNR [13] and we wanted to verify if the same phenomenon exists in the human genome. Diverse gene regions in HG38 were separately analyzed to yield a comparative view of polytract distribution across different types of gene regions. Certain RNA-binding proteins’ binding motifs manifest inherent nucleotide repeats, so the imminent analysis would be able to confirm such internal repeat feature and also possibly indicate nearby or distant repeat features. Finally, more than ten miscellaneous features were included the analysis because they were sequential features with easily attainable genomic locations.

RNA editing events (GRCh37) were merged from REDIportal [36] and DARNED [37], reaching a total of ~4.67m. Because the source databases reported editing events on forward strand and strand indistinguishably (e.g., both A>G and T>C denote the A-to-G class), we grouped all editing events into six classes, which further formed three major categories: A-to-G, C-to-T, and non-canonical Table 2.

Genomic coordinates of 5,967 enhancers in the human genome GRCh37, originally from FANTOM5, were obtained from a published supporting material [38]. These enhancers were found active in at least one of six cancer cell lines. Of all enhancers, we took 4,833 as (cell-line) specific enhancers, and 123 as universal enhancers. Specific enhancers each were active in only one cell line, whereas universal enhancers each were active in four, five, or six cell lines. An intermediate term, non-specific enhancer, was defined for enhancers active in 2~6 cell lines, and these non-specific enhancers totaled 1,134.

GTF (Gene Transfer file) file for the human genome GRCh38 was downloaded from Ensembl (https://uswest.ensembl.org) on 10/17/2018. We kept unique intervals for exons, and organized them into separate files for 47 distinct gene types.

Presumable binding segments of 26 human RNA-binding proteins were downloaded [18] and aligned to GRCh38 via the blastn application [39].

A total of ~2.61m eQTLs (expression quantitative trait loci) in the human genome GRCh37 were obtained from GTEx v7 (p<0.01, Fixed Effect model). 3.6% of these eQTL sites were deletions, for which we considered only the first nucleotides for simplicity.

Mutation Annotation Files for 33 cancer types (GRCh38) were downloaded from Genomic Data Commons (https://gdc.cancer.gov/), which maintains somatic mutation data generated by The Cancer Genomes Atlas (TCGA) project. Merging across all cancer types, we arrived at ~2.62m SNVs (Single-Nucleotide Variations) and ~165k indels.

From the ANNOVAR database [40], we downloaded the gnomAD datasets [19] which stored polymorphic variants among healthy populations. The GRCh38 dataset consisted of ~227k SNVs and ~66.4k indels.

From the UCSC Table browser, we downloaded computationally scanned Transcription Factor Binding Sites (TFBSs) located in GRCh37 [41], which included ~5.49m records.

From the ENCODE project, we downloaded six histone modification datasets that were assayed in human’s muscle of trunk and were accessible in bigWig file formats. These six datasets featured H3K27ac, H3K27me3, H3K4me3, H3K9me3, H3K4me1, and H3K36me3, respectively.

Chromosome coordinates of potentially active LINE-1 (L1) transposons, including completely intact (FLI-L1s) and ORF2-intact (ORF2-L1s) LINE-1s, were downloaded from L1Base [42] (http://l1base.charite.de/l1base.php).

Database euL1db [43] and a published supplementary table [44] collected the confirmed insertion sites (target-site duplication) of human Retrotransposons Insertion Polymorphisms (RIPs). Because these two sources were largely non-overlapping, we merged them to form a single RIP dataset.

Low-interest variants (n = 167,144) recommended as a blacklist for human exome analysis (GRCh38) were obtained from a published supporting material [7].

A list of these miscellaneous genomic features, mapping to the identifiers in Fig 5, is provided in S3 Table.

Miscellaneous statistical analyses

All statistical analyses, including the procedures encompassed in software package Polytrap, were conducted in R environment. We employed the hypergeometric probability model to test for the significant enrichment of MNRs/DNRs/TNRs in enhancers (Fig 2A) based on four key numbers: number of enhancers, number of enhancers embedding polytracts, number of nucleotides in reference genome, and number of nucleotides in reference genome occupied by polytracts. Confidence Interval (95%) of Relative Risk (Fig 2B) was calculated following the standard way [45]. In hierarchical clustering analysis (Fig 3B), distance metric used one minus correlation value and linkage choice was the average method. We employed the over-representation binomial probability model implemented in Polytrap to study RNA-editing events’ enrichment in various polytract species, where the statistical significance was set at nominal p<0.01. Pearson correlation coefficient (r) was calculated and tested for r≠0 when we studied a possible linear correlation relationship between mean length and density of polytracts (S1 Fig).

Supporting information

S1 Table. Comparative summary statistics of polytracts between GRCh38 and GRCh37.

*TNR species are designated with lowercase strings just to imply that the written TNR string actually accommodates multiple TNR motifs. For instance, TNR aat accommodates six repeat motifs: AAT, ATA, TAA, ATT, TAT, and TTA. This phenomenon is different from that of MNR and DNR, where the species name uniquely identifies a repeat motif.

(DOCX)

Click here for additional data file.^{(17.5KB, docx)}

S2 Table. Long DNRs and TNRs present in Polytract dataset yet absent in UCSC Microsatellites.

DNRs and TNRs of 15 or more units present in Polytract dataset yet absent in UCSC Microsatellites (GRCh38).

(TSV)

Click here for additional data file.^{(1.4MB, tsv)}

S3 Table. Genomic features covered in polytract feature enrichment landscapes.

(DOCX)

Click here for additional data file.^{(19.7KB, docx)}

S4 Table. Comparison of Polytrap against related tools.

(DOCX)

Click here for additional data file.^{(27.5KB, docx)}

S1 Fig. Significant linear correlation exists between mean length and density of TNRs across chromosomes.

(TIF)

Click here for additional data file.^{(140.8KB, tif)}

S2 Fig. Distribution of each polytract species across seven genomic region categories.

A, MNR. B, DNR. C, TNR.

(TIF)

Click here for additional data file.^{(421KB, tif)}

S3 Fig. Replicated TNR length distribution and concomitant TNR species clustering.

From the total 1,418,147 TNR tracts, we identified a subset of 39,937 long polytracts that each contained six or more (complete or incomplete) trimer units (i.e., length ≥ 18nt). By repeating the same precedent analytics as performed by Kozlowski et al., we revealed largely the same length distributions for the ten TNR species, and rendered a similar TNR clustering based on a matrix of pairwise Kolmogorov-Smirnov test statistics (S3 Fig). A, polytract length distribution dissected by TNR species. B, heatmap graph showing the Kolmogorov-Smirnov statistics D resulting from pairwise comparisons of polytract lengths among all 10 TNR subtypes.

(TIF)

Click here for additional data file.^{(456.1KB, tif)}

S4 Fig. Relative risk of cancer somatic mutation within polytracts versus whole genome.

TCGA somatic SNVs were significantly over-represented within TNR polytracts but not MNR/DNR polytracts (S4 Fig, A vs. B). Because TCGA mutations were heavily biased to exomes due to experiment design and TNR polytracts have a greater exonic portion than MNR/DNR, we suspected the unique enrichment of somatic SNVs within TNR might just be a trivial reflection of the elevated exonic component of trimer tracts. To interrogate this suspect, we restricted the enrichment analysis to non-coding SNVs only, finding the qualitative distinction between MNR/DNR and TNR persisted for nearly all cancers (S4C Fig). In conclusion, we observed significant enrichment of cancer SNVs within TNR but not within MNR/DNR, even after taking into account the disparate genomic compositions of polytract clades. A, enrichment of SNVs/indels within combined monomer/dimer tracts. B, enrichment of SNVs/indels within trimer tracts. C, enrichment of non-coding SNVs within monomer/dimer or trimer tracts. Asterisk (*) indicates significant enrichment of mutations within polytract regions (p<0.01).

(TIF)

Click here for additional data file.^{(775.8KB, tif)}

S5 Fig. Locational under-representation of certain RNA-editing events in polytracts.

By setting the directionality option of Polytrap to “under,” we switched to investigate the under-representation tendency of three classes of RNA-editing events in polytracts, and found A-to-G events were under-represented in almost all polytract species, excluding A/T, ACC, and hinges. S5 Fig is manifested in analogous format to Fig 4, but the significance asterisk notation (*) in panel A and color shading in panel B designate under-representation rather than over-representation. A, barplot denotes extremity of Relative Risk (RR), with significant (p<0.01) under-representation tendency annotated with asterisk (*). B, heatmap of decreased RR of polytract bearing an RNA-editing event, depicted for combinations between editing event classes and polytract species.

(TIF)

Click here for additional data file.^{(406.6KB, tif)}

Acknowledgments

We thank Luis Nassar and Matthew Speir from UCSC Genomics Institute for assistance in compilation of the tract data hub. We appreciate the four reviewers of our manuscript who raised important concerns and made constructive suggestions throughout rounds of revisions.

Data Availability

All computer code of Polytrap are available from the GitHub database (https://github.com/hui-sheen/polytrap/). Genomic coordinates of polytracts in four model organisms are formatted as a public track for UCSC Genome Browser (http://innovebioinfo.com/Annotation/Polytracts/Polytract.html).

Funding Statement

HY, SN, HK, OO, and YG were supported by the cancer center support grant P30CA118100. This study was supported by the Comprehensive Cancer Center at the University of New Mexico, the Bioinformatics Shared Resources and the Biostatistics Shared Resources at The Comprehensive Cancer Center. None of the funding body was involved in the study design, collection, analysis, interpretation of data, or in writing the manuscript.

References

1.Subramanian S, Mishra RK, Singh L. Genome-wide analysis of microsatellite repeats in humans: their abundance and density in specific genomic regions. Genome Biol. 2003;4(2):R13 10.1186/gb-2003-4-2-r13 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Gymrek M. A genomic view of short tandem repeats. Curr Opin Genet Dev. 2017;44:9–16. 10.1016/j.gde.2017.01.012 . [DOI] [PubMed] [Google Scholar]
3.Pai TW, Chen CM. SSRs as genetic markers in the human genome and their observable relationship to hereditary diseases. Biomark Med. 2016;10(6):563–6. 10.2217/bmm-2016-0094 . [DOI] [PubMed] [Google Scholar]
4.Clark RM, Bhaskar SS, Miyahara M, Dalgliesh GL, Bidichandani SI. Expansion of GAA trinucleotide repeats in mammals. Genomics. 2006;87(1):57–67. 10.1016/j.ygeno.2005.09.006 . [DOI] [PubMed] [Google Scholar]
5.La Spada AR, Taylor JP. Repeat expansion disease: progress and puzzles in disease pathogenesis. Nat Rev Genet. 2010;11(4):247–58. 10.1038/nrg2748 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Gymrek M, Golan D, Rosset S, Erlich Y. lobSTR: A short tandem repeat profiler for personal genomes. Genome Res. 2012;22(6):1154–62. 10.1101/gr.135780.111 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Maffucci P, Bigio B, Rapaport F, Cobat A, Borghesi A, Lopez M, et al. Blacklisting variants common in private cohorts but not in public databases optimizes human exome analysis. Proc Natl Acad Sci U S A. 2019;116(3):950–9. 10.1073/pnas.1808403116 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Willems T, Gymrek M, Highnam G, Genomes Project C, Mittelman D, Erlich Y. The landscape of human STR variation. Genome Res. 2014;24(11):1894–904. 10.1101/gr.177774.114 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Genovese LM, Geraci F, Corrado L, Mangano E, D'Aurizio R, Bordoni R, et al. A Census of Tandemly Repeated Polymorphic Loci in Genic Regions Through the Comparative Integration of Human Genome Assemblies. Front Genet. 2018;9:155 10.3389/fgene.2018.00155 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Benson G. Tandem repeats finder: a program to analyze DNA sequences. Nucleic Acids Res. 1999;27(2):573–80. 10.1093/nar/27.2.573 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Kozlowski P, de Mezer M, Krzyzosiak WJ. Trinucleotide repeats in human genome and exome. Nucleic Acids Res. 2010;38(12):4027–39. 10.1093/nar/gkq127 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Tang H, Kirkness EF, Lippert C, Biggs WH, Fabani M, Guzman E, et al. Profiling of Short-Tandem-Repeat Disease Alleles in 12,632 Human Whole Genomes. Am J Hum Genet. 2017;101(5):700–15. 10.1016/j.ajhg.2017.09.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Yanez-Cuna JO, Arnold CD, Stampfel G, Boryn LM, Gerlach D, Rath M, et al. Dissection of thousands of cell type-specific enhancers identifies dinucleotide repeat motifs as general enhancer features. Genome Res. 2014;24(7):1147–56. 10.1101/gr.169243.113 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Zhao S, Samuels DC, Zhao YY, Guo Y. Tri-allelic heteroplasmies, DNA-RNA differences and their polynucleotide tract associations in the mitochondrial genome. Genomics. 2017. 10.1016/j.ygeno.2017.09.013 . [DOI] [PubMed] [Google Scholar]
15.Guo Y, Yu H, Samuels DC, Yue W, Ness S, Zhao YY. Single-nucleotide variants in human RNA: RNA editing and beyond. Brief Funct Genomics. 2018. 10.1093/bfgp/ely032 . [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Piskol R, Peng Z, Wang J, Li JB. Lack of evidence for existence of noncanonical RNA editing. Nature biotechnology. 2013;31(1):19–20. 10.1038/nbt.2472 . [DOI] [PubMed] [Google Scholar]
17.Guo Y, Dai Y, Yu H, Zhao S, Samuels DC, Shyr Y. Improvements and impacts of GRCh38 human reference on high throughput sequencing data analysis. Genomics. 2017;109(2):83–90. 10.1016/j.ygeno.2017.01.005 . [DOI] [PubMed] [Google Scholar]
18.Yu H, Wang J, Sheng Q, Liu Q, Shyr Y. beRBP: binding estimation for human RNA-binding proteins. Nucleic Acids Res. 2019;47(5):e26 10.1093/nar/gky1294 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Lek M, Karczewski KJ, Minikel EV, Samocha KE, Banks E, Fennell T, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536(7616):285–91. 10.1038/nature19057 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Gao GF, Parker JS, Reynolds SM, Silva TC, Wang LB, Zhou W, et al. Before and After: Comparison of Legacy and Harmonized TCGA Genomic Data Commons' Data. Cell Syst. 2019;9(1):24–34 e10. 10.1016/j.cels.2019.06.006 . [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Liu S, Hou W, Sun T, Xu Y, Li P, Yue B, et al. Genome-wide mining and comparative analysis of microsatellites in three macaque species. Mol Genet Genomics. 2017;292(3):537–50. 10.1007/s00438-017-1289-1 . [DOI] [PubMed] [Google Scholar]
22.Merkel A, Gemmell N. Detecting short tandem repeats from genome data: opening the software black box. Brief Bioinform. 2008;9(5):355–66. 10.1093/bib/bbn028 . [DOI] [PubMed] [Google Scholar]
23.Pandey M, Kumar R, Srivastava P, Agarwal S, Srivastava S, Nagpure NS, et al. WGSSAT: A High-Throughput Computational Pipeline for Mining and Annotation of SSR Markers From Whole Genomes. J Hered. 2018;109(3):339–43. 10.1093/jhered/esx075 . [DOI] [PubMed] [Google Scholar]
24.Nie X, Huang C, You C, Li W, Zhao W, Shen C, et al. Genome-wide SSR-based association mapping for fiber quality in nation-wide upland cotton inbreed cultivars in China. BMC Genomics. 2016;17:352 10.1186/s12864-016-2662-x [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Raney BJ, Dreszer TR, Barber GP, Clawson H, Fujita PA, Wang T, et al. Track data hubs enable visualization of user-defined genome-wide annotations on the UCSC Genome Browser. Bioinformatics. 2014;30(7):1003–5. 10.1093/bioinformatics/btt637 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Willems T, Zielinski D, Yuan J, Gordon A, Gymrek M, Erlich Y. Genome-wide profiling of heritable and de novo STR variations. Nat Methods. 2017;14(6):590–2. 10.1038/nmeth.4267 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Highnam G, Franck C, Martin A, Stephens C, Puthige A, Mittelman D. Accurate human microsatellite genotypes from high-throughput resequencing data using informed error profiles. Nucleic Acids Res. 2013;41(1):e32 10.1093/nar/gks981 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Sheng Q, Yu H, Oyebamiji O, Wang J, Chen D, Ness S, et al. AnnoGen: Annotating Genomewide Pragmatic Features. Bioinformatics. 2020. 10.1093/bioinformatics/btaa027 . [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Lin W, Piskol R, Tan MH, Li JB. Comment on "Widespread RNA and DNA Sequence Differences in the Human Transcriptome". Science. 2012;335(6074). 10.1126/science.1210624 WOS:000301531600026. [DOI] [PubMed] [Google Scholar]
30.Schrider DR, Gout JF, Hahn MW. Very Few RNA and DNA Sequence Differences in the Human Transcriptome. Plos One. 2011;6(10). ARTN e25842 10.1371/journal.pone.0025842 WOS:000295976000044. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Lee JH, Ang JK, Xiao X. Analysis and design of RNA sequencing experiments for identifying RNA editing and other single-nucleotide variants. RNA. 2013;19(6):725–32. 10.1261/rna.037903.112 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Liu Z, Zhang J. Human C-to-U Coding RNA Editing Is Largely Nonadaptive. Mol Biol Evol. 2018;35(4):963–9. 10.1093/molbev/msy011 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.McLean CY, Bristor D, Hiller M, Clarke SL, Schaar BT, Lowe CB, et al. GREAT improves functional interpretation of cis-regulatory regions. Nature biotechnology. 2010;28(5):495–501. 10.1038/nbt.1630 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Otlu B, Firtina C, Keles S, Tastan O. GLANET: genomic loci annotation and enrichment tool. Bioinformatics. 2017;33(18):2818–28. 10.1093/bioinformatics/btx326 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 2012;40(Database issue):D930–4. 10.1093/nar/gkr917 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Picardi E, D'Erchia AM, Lo Giudice C, Pesole G. REDIportal: a comprehensive database of A-to-I RNA editing events in humans. Nucleic Acids Res. 2017;45(D1):D750–D7. 10.1093/nar/gkw767 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Kiran A, Baranov PV. DARNED: a DAtabase of RNa EDiting in humans. Bioinformatics. 2010;26(14):1772–6. 10.1093/bioinformatics/btq285 . [DOI] [PubMed] [Google Scholar]
38.Li Y, Shi W, Wasserman WW. Genome-wide prediction of cis-regulatory regions using supervised deep learning methods. BMC Bioinformatics. 2018;19(1):202 10.1186/s12859-018-2187-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Morgulis A, Coulouris G, Raytselis Y, Madden TL, Agarwala R, Schaffer AA. Database indexing for production MegaBLAST searches. Bioinformatics. 2008;24(16):1757–64. 10.1093/bioinformatics/btn322 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38(16):e164 10.1093/nar/gkq603 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Liu GE, Weirauch MT, Van Tassell CP, Li RW, Sonstegard TS, Matukumalli LK, et al. Identification of conserved regulatory elements in mammalian promoter regions: a case study using the PCK1 promoter. Genomics Proteomics Bioinformatics. 2008;6(3–4):129–43. 10.1016/S1672-0229(09)60001-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Penzkofer T, Jager M, Figlerowicz M, Badge R, Mundlos S, Robinson PN, et al. L1Base 2: more retrotransposition-active LINE-1s, more mammalian genomes. Nucleic Acids Res. 2017;45(D1):D68–D73. 10.1093/nar/gkw925 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Mir AA, Philippe C, Cristofari G. euL1db: the European database of L1HS retrotransposon insertions in humans. Nucleic Acids Res. 2015;43(Database issue):D43–7. 10.1093/nar/gku1043 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Yu Q, Zhang W, Zhang X, Zeng Y, Wang Y, Wang Y, et al. Population-wide sampling of retrotransposon insertion polymorphisms using deep sequencing and efficient detection. Gigascience. 2017;6(9):1–11. 10.1093/gigascience/gix066 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Morris JA, Gardner MJ. Calculating confidence intervals for relative risks (odds ratios) and standardised ratios and rates. Statistics in Medicine. 1988;296:1313–6. [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS Comput Biol. doi: 10.1371/journal.pcbi.1007968.r001

Decision Letter 0

Lilia M Iakoucheva, William Stafford Noble

7 Jan 2020

Dear Dr Yu,

Thank you very much for submitting your manuscript 'Non-canonical RNA-DNA differences and other human genomic features are associated with very short tandem repeats' for review by PLOS Computational Biology. Your manuscript has been fully evaluated by the PLOS Computational Biology editorial team and in this case also by independent peer reviewers. The reviewers appreciated the attention to an important problem, but raised some substantial concerns about the manuscript as it currently stands. While your manuscript cannot be accepted in its present form, we are willing to consider a revised version in which the issues raised by the reviewers have been adequately addressed. We cannot, of course, promise publication at that time.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

Your revisions should address the specific points made by each reviewer. Please return the revised version within the next 60 days. If you anticipate any delay in its return, we ask that you let us know the expected resubmission date by email at ploscompbiol@plos.org. Revised manuscripts received beyond 60 days may require evaluation and peer review similar to that applied to newly submitted manuscripts.

In addition, when you are ready to resubmit, please be prepared to provide the following:

(1) A detailed list of your responses to the review comments and the changes you have made in the manuscript. We require a file of this nature before your manuscript is passed back to the editors.

(2) A copy of your manuscript with the changes highlighted (encouraged). We encourage authors, if possible to show clearly where changes have been made to their manuscript e.g. by highlighting text.

(3) A striking still image to accompany your article (optional). If the image is judged to be suitable by the editors, it may be featured on our website and might be chosen as the issue image for that month. These square, high-quality images should be accompanied by a short caption. Please note as well that there should be no copyright restrictions on the use of the image, so that it can be published under the Open-Access license and be subject only to appropriate attribution.

Before you resubmit your manuscript, please consult our Submission Checklist to ensure your manuscript is formatted correctly for PLOS Computational Biology: http://www.ploscompbiol.org/static/checklist.action. Some key points to remember are:

- Figures uploaded separately as TIFF or EPS files (if you wish, your figures may remain in your main manuscript file in addition).

- Supporting Information uploaded as separate files, titled Dataset, Figure, Table, Text, Protocol, Audio, or Video.

- Funding information in the 'Financial Disclosure' box in the online system.

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see here.

We are sorry that we cannot be more positive about your manuscript at this stage, but if you have any concerns or questions, please do not hesitate to contact us.

Sincerely,

Lilia M. Iakoucheva, Ph.D.

Associate Editor

PLOS Computational Biology

William Noble

Deputy Editor

PLOS Computational Biology

A link appears below if there are any accompanying review attachments. If you believe any reviews to be missing, please contact ploscompbiol@plos.org immediately:

[LINK]

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: The authors of the reviewed article (PCOMPBIOL-D-19-01709) aimed to catalog and characterize the mono- di- and tri-nucleotide tracts present in the reference human genome and looked for the associations of the tracts with numerous genomic features, e.g., RNA editing sites. Although, as admitted by the authors, this type of study is not completely new, however previous studies usually were focused on a particular type of repeats or particular genomic regions and often were performed in pre- or early-genomic era. Therefore, systematic complex descriptive analysis of all types (1-3 nt) of microsatellites still could have been rationalizing the study. Additionally, the authors have extended the analysis to relatively short repeat tracts (minimal length of 6, 6, and 9 nt for mono- di- and tri-nucleotide tracts, respectively) that are extremely abundant in the genome and which analysis/inclusion could have been a real novelty of the study. The study is well justified.

Nevertheless, in my opinion, the aspect of descriptive characterization of the repeats is not fully explored and is not informatively presented. The manuscript has also several serious drawbacks that would have to be clarified/corrected before considering the manuscript for publication.

1. The redundant repeats should be merged as it is done for triplets. There is no need to distinguish complementary tracts (e.g., A and T or GA and TC) unless located and considered in the context of the direction of a gene or other genomic elements (it is not the case of the present version of the study). Otherwise, the genomic orientation of repeat tracts is random, for example, please see almost identical statistics for complementary tracts in Appendix 1. The distinguishing the complementary tracts makes a lot of analyses redundant and unnecessarily confusing.

2. The authors very inconsistently use names, descriptions for particular repeat tracts. The single/specific name should be selected for a particular tract at the beginning and then it should be used consequently throughout the entire manuscript, including pictures and supplementary materials. For example, in the Introduction the authors abbreviate trinucleotide repeats as TNRs but than use different terms, including tri-nucleotide tracts, Tri-nucleotide tracts, tri-tracts, trimers, tri, tri-, as well as again explain and use the abbreviation TNR. I would suggest to consequently to trinucleotide repeats (TNR) abbreviate mononucleotide and dinucleotide repeats (eg. MNR and DNR, respectively) and use them consequently in the manuscript.

3. Results, 2nd paragraph: “ We identified the same three most exome-prominent tri-tracts as previous researchers did (9): GAC, CGG, and CAG, but presenting them in a different exome-prominence order (GAC > CGG > 121 CAG, rather than CGG>CAG>GAC (9)).” In the previous study (9) the 3rd most exome prominent TNR was AGG (not GAC). GAC was the 3rd most exome overrepresented (see the difference between Fig. 1A and Fig. 1B in (9)). GAC was the least frequent TNR, with only 16 occurrences in the genome, including 3 in exons.

4. Results, 4th paragraph: It should be clarified whether density is a number of tracks per Mb or a number of nucleotides in tracks per Mb. It is critical for understanding the last sentence in the paragraph. To this paragraph but also generally to the study – due to the high overrepresentation of shortest tracts (extremally not normal distribution) the differences between tracks lengths are very small and therefore analyses/comparisons of means or even medians of the repeat lengths are of little value (e.g. differences in repeats lengths between chromosomes in Appendix 1.

5. Results, 5th paragraph, the part starting with “An intuitively plausible ...” to the end of the paragraph. I got lost in this part. Despite that I am quite familiar with the subject, it took me quite a long time to comprehend this paragraph. I would suggest to reconsider and to rewrite the paragraph.

6. Results: I do not understand the sentence ‘AAT is the most prevalent TNR species (Fig. 1E), bearing a potential replication regulatory role in partnership of 4′,6‐diamidine‐2‐phenylindole (DAPI) (15)’.

7. Results, the paragraph starting with line 196: As mentioned above, as differences in TNRs length are very small the analysis described in the paragraph is of little value. It would be more interesting and more informative to compare the frequency of TNRs occurrence in particular genomic/gene regions. It should be compared separately for each TNR type, and in genes also with consideration of genes orientation. Same for other types of repeats.

8. Resalts, the sentence starting with line 223: Please explain the better.

9. Results, line 197: Why one-tiled test? Please justify. I think it should be the two-tiled test.

10. It is not always clear to me what exactly was used as a background for calculation of expected frequency or cooccurrence of repeats with particular genomic features. It is important because some repeats may associate with genomic regions enriched with the tested feature and it may lead to a spurious association. For example, it is not clarified what type of cancer somatic mutations were extracted from TCGA (whole-genome?, total whole-exome including exon flanking sequences?, only exon sequences including UTRs?, or exon ORFs excluding UTR?) and in consequence what was used as a target and background sequence for analysis. It is known that particular repeat types associates with these subregions, therefore, some comparisons may lead to false-positive spurious associations,

11. Materials and Methods, ‘in human reference genomes GRCh37 and GRCh38, respectively’: What does it mean? In which assembly which repeats types have been analyzed? Why two assemblies were used?

12. Materials and Methods, ‘H1 (alternative hypothesis) - the concerned genomic feature is distributed in polytract territory at a frequency higher than random expectation.’ Why only frequency higher than expected were considered as H1? In my opinion, both over and underrepresentations should be considered and accordingly, two-sided tests should be used. Please explain.

13. Appendix 1 and corresponding text fragments: It would be more interesting to see comparisons of particular repeat types occurrence in particular genome/gene subregions (5’UTR, 3’UTR, ORF, introns, intergenic regions) instead of in different chromosomes, in which occurrence mostly depends on chromosome length and therefore frequency or density do not differ substantially. The exception may be chromosome Y, X, and MT. Autosomes may be combined.

14. The legend in Figure 1B: What is it ‘exome splicing’? What about the coding sequence subregion? It is not indicated.

15. Figure 3B and corresponding fragment of Results: Heatmap and dendrogram do not show unique distributions of CGG, ACC, and ATT. ATT show very similar distribution with AGG and quite similar distribution (same clade) with all other but CGG and ACC TNRs. The interpretation of Fig 3B should be reconsidered. Also, there is something wrong with the bitmap. It shows that the comparison of AAT with AAT (with itself) gives D-value different from 0. The same is for AGG. The same problem is with the bitmap in the supplementary Figure (Appendix 6).

16. Figure 4A and corresponding fragments of Results. It is shown that RNA editing events are not only enriched but also underrepresented in some repeat types (eg. A2G is underrepresented in dinucleotide tracts and TNRs). The fold of these changes is similar to some positive changes but the significance of these changes is not indicated. The decreases in RNA-editing events are not commented in the text. Please explain/comment on the lack of significance of the negative changes and the fact that repeats generate both increases and decreases of RNA-editing events or artifacts.

17. Figure 5: It is very surprising that the significant associations ordered according to p-values correlate perfectly between repeat classes, i.e., the same order of genomic features in mono- di, and tri-nucleotide repeats. It may suggest spurious associations. I would suggest double-checking.

18. Figure 5: It would be interesting to see also fold changes (not only p-values) for particular features. To see whether an association is positive or negative. Also, it would be interesting to see how the particular subtypes of mono-, di-, and tri-nucleotide tracts associate with the genomic features.

19. The list of identified tracts, their type, genomic coordinates, and potentially associated genomic features should be always available for potentially interested readers. The UCSC custom tracts for the particular repeats are deposited on private google drive. I have bad experiences with “upon request” or private repositories. They usually do not work shortly after publication. Also, there is no need to mention the custom tracts for other organisms in the abstract if they are not analyzed or discussed in the paper.

20. Appendix 1: The title of the subtable (spreadsheet) ‘genomeOccupancy’ should be rather “chromosomeOccupancy”.

Reviewer #2: This paper presents an updated census of very short tandem repeats (polytracts) particularly in human reference genome GRCh38. Further, authors report their findings of enrichment and association analysis of polytracts (mainly monomers, dimers and trimers) and RNA-edit events both canonical and non-canonical with respect to various genomic features. The authors also provide polytrap, a software tool to perform such association analysis.

Overall, these findings could be of interest to a broad audience in the scientific community. However, it would be a bit of a stretch to confirm such findings when the methodology is not well presented. I would highly recommended the authors to rewrite their materials and methods to explicitly present their approach and how they conducted their analysis. In the following, I indicate a few points that should be addressed:

• There are fundamentally major weaknesses in this manuscript:

o Authors claim that previous analysis of polytracts are outdated and imprecise. I find the argument of improved sequencing technology and better curation of the current version GRCh38 reference genome more plausible, but beside that, the authors fail to present evidence of how their analysis has led to improved detection and characterization.

o Authors do not provide the theoretical framework behind the identification, detection and categorization of polytracts. Authors relied on BSgenome, a Bioconductor package. I find the details in Methods and Materials insufficient in terms of polytracts detection and identification. Authors must include the underlying logical framework and methodology used to infer polytracts from a DNA sequence (reference genome). Authors should indicate whether the method considered leads to improved detection of polytracts especially monomers and dimers that have not been widely reported.

o The paper as it is now has no details or description of the statistical analysis for the results presented. Authors should include a section describing and discussing the statistical analysis and statistical tests performed.

o Authors present polytrap and yet have not described its algorithmic idea and how it works. I encourage the authors to include such information as well.

• The Authors use the binomial distribution and clearly express the hypotheses they wish to test. The authors should explain why they considered this distribution and any underlying assumptions made and how this relates to the downstream association analysis.

o How do you define and determine the tract territory?

o L387-L392: The singleton mode and the multiplex mode, it should be helpful to visually illustrate the different modes if possible.

o Authors use terms like tract number, density, tract volume and genome occupancy and yet do not explain them in the main text nor in the supplementary materials [Supp. S1].

• Location of miscellaneous genomic features

o Authors should briefly discuss why they considered the specified genomic features and discuss briefly their biological importance for instance with respect to RDDs

o It appears to me that the authors only listed materials gathered for their analysis. They fail to present how they actually conducted and performed their analysis. You have collected data from various sources with potentially different genomic coordinates (GRCh37 vs GRCh38, for instance) and yet you do not mention how all that information was integrated and used.

o Overall, the materials and method part of the paper needed to be significantly improved

• Minor issues:

o Provide citation L151

o L215: For consistency, authors should report their RR as done at L223 also to avoid confusion with citation notations.

o L323-L329: I would encourage authors to discuss and put forth the argument of why analysis of A2G editing events are needed in the context of host polytract and probably mention their current manuscript in preparation rather than reporting their findings.

o Proofread the text

o The figures could be a of a better quality and higher resolution

Reviewer #3: This article is well written and thoroughly tries to review/census monomers, dimers, and trinucleotide (tandem) repeats found in GRCh38, by correlating their distribution and composition to specific genomic features and loci.

However, as a first observation, no comparison is made with, for example, simple repeats (UCSC track, last updated on March 2019) and works associating them with human diseases and other genomic features, as well as with the mentioned 700k catalog by Willems et al. and other catalogues and censuses that can be easily retrieved from online (e.g. https://www.frontiersin.org/articles/10.3389/fgene.2018.00155/full).

More importantly, no accurate details and definitions are given for the discovery of such (tandem) repeats, while these could clarify some of the statistical findings and help identifying potential biases with the analysis; sophisticated software and methodologies are nevertheless available from other works, which allow for a flexible and robust tandem repeat finding (e.g. TRF and more recent tools). Reverse complement is also only considered for trinucleotides, while the analysis should be comparable between all three types for some statistics.

Furthermore, the fact that the main differences between DNA and RNA are found inside or adjacent to monomer repeats might easily suggest that these come either from sequencing errors, alignment consensus errors, or possible (low-frequency) polymorphisms. Indeed, monomers, especially long monomers, can be hardly sequenced well and are known to create several artefacts and analysis issues both in sequencing, pre-processing, and post-processing phases, as well as are more prone to indels of a different nature (than the repeat itself) than the other two types. Not surprisingly, significant differences in the distribution of somatic mutations in cancer genomes is noted for both them and trinucleotides (i.e. the size of codons). I would thus suggest to avoid, at least for the moment, mainly focusing on these discoveries both in the title and the abstract.

Finally, no information is available on the created software, as well as on its memory footprint and general performance.

Reviewer #4: Based on the new human reference genome GRCh38, this manuscript studied the genome-wide short tandem repeats (STR) characteristics, in particular the monomer, dimer, and TNR and the implication on RNA-DNA differences and other genome features, which has provided new insight into our understanding of the impact of STR on genome structure and function. In addition, they developed a software to characterize STRs in the genomes, which will be helpful for other researchers. However, some questions are not reasonable and needed to be clarified.

Major comments:

1.The authors used infrastructure packages from Bioconductor (www.bioconductor.org;

DOI: 10.18129/B9.bioc.BSgenome) to identify monomer, dimer, and TNR in human reference genomes，and found the abundance of very short tandem repeats in the new reference genome was much higher than those in the previous reference genome. I am wondering why the authors did not apply some traditional methods such as RepeatSeq or lobSTR to recognize STRs to the reference genome? The authors should justify their methodology and figure out whether the new version of reference genome or the data procession cause such differences on STR abundance. The authors also can check this literature to find better way to deal with STRs.

Willems T, Zielinski D, Yuan J, Gordon A, Gymrek M, Erlich. Genome-wide profiling of heritable and de novo STR variations. Nature Methods. 2017, 14, 590–592.

2.Considering a large number of previous studies on human genome STRs and their characteristics, it is suggested that the author supplement the differences between GRCH38 and previous version, and add the comparison of their results with other studies based on the old version of reference genome.

3.Although the detail analysis has been conducted on the GRCH38 reference genome in the present study, it is only based on one genome. As known for us, short tandem repeat sequences show high polymorphism between genome and genome. How can an analysis based on a single genome reflect the real characteristics of STRs in human or how high is the reliability? The authors should clarify.

4.Based on the new reference genome, the authors analyzed the RNA editing sites associated with STRs, which is one of the highlights of this manuscript. It is suggested that the authors should strengthen this in the Results, in particular they should add some comments in the Discussion to highlight the significance of STRs on RNA editing events in combination with relevant literatures on human RNA editing in recent years.

5.The manuscript developed a software to assist with custom association analysis of polytracts with respect to any localizable genomic feature. Unfortunately, trying to assist the Polytrap is failed to me. Furthermore, the link provided by the manuscript about genomic coordinates of polytracts in nine model organisms is also unavailable. Please provide validated way for using and evaluating the Polytrap. And i would like to know if the Polytrap could be used for population scale analysis.

6.There are many recent literatures about STRs or RNA editing in human genome. In the discussion part, the authors seldom used the literatures and compared their results with the previous studies. Many sentences in the discussion are repeat the results, thus suggest the authors to improve.

Minor comments:

In the Abstract and Keywords, the authors used C-to-U RNA-editing, however in the text part, they used A2T or C2T. This should be revised.

**********

Have all data underlying the figures and results presented in the manuscript been provided?

Large-scale datasets should be made available via a public repository as described in the PLOS Computational Biology data availability policy, and numerical data that underlies graphs or summary statistics should be provided in spreadsheet form as supporting information.

Reviewer #1: No: The lists of identified mono-, di-, and tri-nucleotide tracts provided as custom track files for UCSC Genome Browser are deposited (linked) only on private Google Drive. These lists are very important result of the study and should be deposited in public database or if possible as supplementary materials.

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: No: the polytrap is unaccessable

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Reviewer #4: No

PLoS Comput Biol. 2020 Jun 8;16(6):e1007968. doi: 10.1371/journal.pcbi.1007968.r002

Author response to Decision Letter 0

25 Feb 2020

Attachment

Submitted filename: Polytrap_Resp2referees.docx

Click here for additional data file.^{(824.5KB, docx)}

PLoS Comput Biol. doi: 10.1371/journal.pcbi.1007968.r003

Decision Letter 1

Lilia M Iakoucheva, William Stafford Noble

27 Mar 2020

Dear Dr. Yu,

Thank you very much for submitting your manuscript "Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats" for consideration at PLOS Computational Biology. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. The reviewers appreciated the attention to an important topic. Based on the reviews, we are likely to accept this manuscript for publication, providing that you modify the manuscript according to the review recommendations.

Please prepare and submit your revised manuscript within 30 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to all review comments, and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Thank you again for your submission to our journal. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Lilia M. Iakoucheva, Ph.D.

Associate Editor

PLOS Computational Biology

William Noble

Deputy Editor

PLOS Computational Biology

***********************

A link appears below if there are any accompanying review attachments. If you believe any reviews to be missing, please contact ploscompbiol@plos.org immediately:

[LINK]

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: Detailed comments:

1. L149-L150: The high GC content of hinges connecting AT-reach tracts (i.e., A/T_A/T or TA_TA) is not surprising. I would say that it is rather expected, as the presence of either A or T in the hinge positions would, in most cases, extended the tracts and would not be counted. The same I think applies to GC-reach tracts that, I expect, are mostly separated by AT-reach hinges.

2. Table S2: I do not know whether I understand correctly the column ‘Polytract As Microsatellite’. I would expect that the values in the column should be equal (should not be higher) with the values in the ‘Microsatellite’ column. It is the case for most but not for all polytracts. Please explain or correct it.

3. Figure 1E and Figure S2: What is up_down_stream? How it is defined?

4. Maybe I am too pedantic but why the authors sometimes use a lower-case and sometime upper-case letter for different polytracts (e.g., TA or ta) or why sometimes use for DNR e.g., CT and in other places CT/GA. For readers, especially when someone browses the article quickly such inconsistencies may be confusing.

5. Figure S2 and corresponding text: It is worth to note that GC DNRs are strongly enriched in 5’UTR, exonic, and up_down_stream regions.

6. Does ‘exonic’ mean protein-coding part of exons or exons including UTRs?.

7. According to Figure 3B (bitmap), CGG TNRs are more similar to AATs that to AACs. The fragment of text commenting Figure 3 should be reconsidered.

8. Fragment encompassing L189~L191: Figure 3A does not show repeats “6U or more” but ‘5U or more’ (15+).

Reviewer #2: The authors of the revised manuscript “Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats” have made substantial improvement over the original manuscript. They have satisfactorily addressed most of the concerns and questions I raised earlier, but I still find the article missing an important component: description of performed statistical analyses. Authors performed various statistical tests as illustrated throughout the manuscript mainly in the Results section and shown in the Figures. Though the fundamental statistical analysis incorporated in the software polytrap is well described in the Materials and methods, a detailed description of the performed ad hoc statistical analyses is missing [including which statistical softwares were used]. In addition, it would be necessary to know for instance based on polytrap output, what is the cut off p-value for statistical significance in enrichment analysis.

In addition, the figures especially Figures in the main manuscript are not of publishable quality [when you print them you can hardly read the figure annotations and labels]. In addition, in Figure 6, N is the total number of nucleotides in the genome, but it is indicated as if it is the total number of nucleotides excluding the total number of nucleotides occupied by polytracts. By your definition, N in Fig. 6 should be illustrated by blue color + grey color.

Minor typos on lines L412 [I suppose you meant hypergeometric model instead of hypegenometric model], 442-443 [the word “instead” makes the sentence reads a little bit contradictory to the previous sentence].

Reviewer #3: The article has been reviewed fairly well and most issues have been addressed.

Although an exact tandem repeat finder may hide useful approximate tandem repeats or shorten/break up longer chains, and therefore somehow mislead the statistics you have found, your catalog could help reporting new locations/repeats (absurdly) overlooked before (e.g. in UCSC Simple/Microsatellite Repeats). In this regard, have you checked that your new locations/repeats are correct? Could you justify why they don't appear in the UCSC catalog? Can you confirm that the new locations/repeats are not present in catalogs like the 700k by Willems et al.?

In addition, I would strongly suggest to emphasize earlier, rather than in the Discussion section, that the main differences between DNA and RNA found within or adjacent to monomer repeats may have been caused by artifacts (e.g. sequencing errors, alignment consent errors) or potential (low-frequency) polymorphisms, and similarly for what is reported for somatic mutations in cancer genomes for both mononucleotides and trinucleotides (i.e. the size of codons).

Reviewer #4: The authors have almost answered questions concerned by me. They added related information in methods part, and rewrited the disscussion. The new version looks much better than last version. I suggest to accept the manuscript to publish on the Plos Computaional Biology.

**********

Have all data underlying the figures and results presented in the manuscript been provided?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: Yes: Davide Verzotto

Reviewer #4: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example in PLOS Biology see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, PLOS recommends that you deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/ploscompbiol/s/submission-guidelines#loc-materials-and-methods

PLoS Comput Biol. 2020 Jun 8;16(6):e1007968. doi: 10.1371/journal.pcbi.1007968.r004

Author response to Decision Letter 1

23 Apr 2020

Attachment

Submitted filename: Polytract.p2pResponse.R2.docx

Click here for additional data file.^{(810.4KB, docx)}

PLoS Comput Biol. doi: 10.1371/journal.pcbi.1007968.r005

Decision Letter 2

Lilia M Iakoucheva, William Stafford Noble

19 May 2020

Dear Dr. Yu,

We are pleased to inform you that your manuscript 'Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats' has been provisionally accepted for publication in PLOS Computational Biology.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Computational Biology.

Best regards,

Lilia M. Iakoucheva, Ph.D.

Associate Editor

PLOS Computational Biology

William Noble

Deputy Editor

PLOS Computational Biology

***********************************************************

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: No more comments.

Reviewer #2: The authors have made satisfactorily and significant improvements compared to the last manuscript.

**********

Have all data underlying the figures and results presented in the manuscript been provided?

Reviewer #1: Yes

Reviewer #2: Yes

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Piotr Kozlowski, Institute of Bioorganic Chemistry, Polish Academy of Sciences, Poznan, Poland

Reviewer #2: No

PLoS Comput Biol. doi: 10.1371/journal.pcbi.1007968.r006

Acceptance letter

Lilia M Iakoucheva, William Stafford Noble

2 Jun 2020

PCOMPBIOL-D-19-01709R2

Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats

Dear Dr Yu,

I am pleased to inform you that your manuscript has been formally accepted for publication in PLOS Computational Biology. Your manuscript is now with our production department and you will be notified of the publication date in due course.

The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript.

Soon after your final files are uploaded, unless you have opted out, the early version of your manuscript will be published online. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers.

Thank you again for supporting PLOS Computational Biology and open-access publishing. We are looking forward to publishing your work!

With kind regards,

Matt Lyles

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Table. Comparative summary statistics of polytracts between GRCh38 and GRCh37.

(DOCX)

Click here for additional data file.^{(17.5KB, docx)}

S2 Table. Long DNRs and TNRs present in Polytract dataset yet absent in UCSC Microsatellites.

DNRs and TNRs of 15 or more units present in Polytract dataset yet absent in UCSC Microsatellites (GRCh38).

(TSV)

Click here for additional data file.^{(1.4MB, tsv)}

S3 Table. Genomic features covered in polytract feature enrichment landscapes.

(DOCX)

Click here for additional data file.^{(19.7KB, docx)}

S4 Table. Comparison of Polytrap against related tools.

(DOCX)

Click here for additional data file.^{(27.5KB, docx)}

S1 Fig. Significant linear correlation exists between mean length and density of TNRs across chromosomes.

(TIF)

Click here for additional data file.^{(140.8KB, tif)}

S2 Fig. Distribution of each polytract species across seven genomic region categories.

A, MNR. B, DNR. C, TNR.

(TIF)

Click here for additional data file.^{(421KB, tif)}

S3 Fig. Replicated TNR length distribution and concomitant TNR species clustering.

(TIF)

Click here for additional data file.^{(456.1KB, tif)}

S4 Fig. Relative risk of cancer somatic mutation within polytracts versus whole genome.

(TIF)

Click here for additional data file.^{(775.8KB, tif)}

S5 Fig. Locational under-representation of certain RNA-editing events in polytracts.

(TIF)

Click here for additional data file.^{(406.6KB, tif)}

Attachment

Submitted filename: Polytrap_Resp2referees.docx

Click here for additional data file.^{(824.5KB, docx)}

Attachment

Submitted filename: Polytract.p2pResponse.R2.docx

Click here for additional data file.^{(810.4KB, docx)}

Data Availability Statement

[pcbi.1007968.ref001] 1.Subramanian S, Mishra RK, Singh L. Genome-wide analysis of microsatellite repeats in humans: their abundance and density in specific genomic regions. Genome Biol. 2003;4(2):R13 10.1186/gb-2003-4-2-r13 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref002] 2.Gymrek M. A genomic view of short tandem repeats. Curr Opin Genet Dev. 2017;44:9–16. 10.1016/j.gde.2017.01.012 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref003] 3.Pai TW, Chen CM. SSRs as genetic markers in the human genome and their observable relationship to hereditary diseases. Biomark Med. 2016;10(6):563–6. 10.2217/bmm-2016-0094 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref004] 4.Clark RM, Bhaskar SS, Miyahara M, Dalgliesh GL, Bidichandani SI. Expansion of GAA trinucleotide repeats in mammals. Genomics. 2006;87(1):57–67. 10.1016/j.ygeno.2005.09.006 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref005] 5.La Spada AR, Taylor JP. Repeat expansion disease: progress and puzzles in disease pathogenesis. Nat Rev Genet. 2010;11(4):247–58. 10.1038/nrg2748 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref006] 6.Gymrek M, Golan D, Rosset S, Erlich Y. lobSTR: A short tandem repeat profiler for personal genomes. Genome Res. 2012;22(6):1154–62. 10.1101/gr.135780.111 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref007] 7.Maffucci P, Bigio B, Rapaport F, Cobat A, Borghesi A, Lopez M, et al. Blacklisting variants common in private cohorts but not in public databases optimizes human exome analysis. Proc Natl Acad Sci U S A. 2019;116(3):950–9. 10.1073/pnas.1808403116 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref008] 8.Willems T, Gymrek M, Highnam G, Genomes Project C, Mittelman D, Erlich Y. The landscape of human STR variation. Genome Res. 2014;24(11):1894–904. 10.1101/gr.177774.114 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref009] 9.Genovese LM, Geraci F, Corrado L, Mangano E, D'Aurizio R, Bordoni R, et al. A Census of Tandemly Repeated Polymorphic Loci in Genic Regions Through the Comparative Integration of Human Genome Assemblies. Front Genet. 2018;9:155 10.3389/fgene.2018.00155 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref010] 10.Benson G. Tandem repeats finder: a program to analyze DNA sequences. Nucleic Acids Res. 1999;27(2):573–80. 10.1093/nar/27.2.573 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref011] 11.Kozlowski P, de Mezer M, Krzyzosiak WJ. Trinucleotide repeats in human genome and exome. Nucleic Acids Res. 2010;38(12):4027–39. 10.1093/nar/gkq127 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref012] 12.Tang H, Kirkness EF, Lippert C, Biggs WH, Fabani M, Guzman E, et al. Profiling of Short-Tandem-Repeat Disease Alleles in 12,632 Human Whole Genomes. Am J Hum Genet. 2017;101(5):700–15. 10.1016/j.ajhg.2017.09.013 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref013] 13.Yanez-Cuna JO, Arnold CD, Stampfel G, Boryn LM, Gerlach D, Rath M, et al. Dissection of thousands of cell type-specific enhancers identifies dinucleotide repeat motifs as general enhancer features. Genome Res. 2014;24(7):1147–56. 10.1101/gr.169243.113 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref014] 14.Zhao S, Samuels DC, Zhao YY, Guo Y. Tri-allelic heteroplasmies, DNA-RNA differences and their polynucleotide tract associations in the mitochondrial genome. Genomics. 2017. 10.1016/j.ygeno.2017.09.013 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref015] 15.Guo Y, Yu H, Samuels DC, Yue W, Ness S, Zhao YY. Single-nucleotide variants in human RNA: RNA editing and beyond. Brief Funct Genomics. 2018. 10.1093/bfgp/ely032 . [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref016] 16.Piskol R, Peng Z, Wang J, Li JB. Lack of evidence for existence of noncanonical RNA editing. Nature biotechnology. 2013;31(1):19–20. 10.1038/nbt.2472 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref017] 17.Guo Y, Dai Y, Yu H, Zhao S, Samuels DC, Shyr Y. Improvements and impacts of GRCh38 human reference on high throughput sequencing data analysis. Genomics. 2017;109(2):83–90. 10.1016/j.ygeno.2017.01.005 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref018] 18.Yu H, Wang J, Sheng Q, Liu Q, Shyr Y. beRBP: binding estimation for human RNA-binding proteins. Nucleic Acids Res. 2019;47(5):e26 10.1093/nar/gky1294 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref019] 19.Lek M, Karczewski KJ, Minikel EV, Samocha KE, Banks E, Fennell T, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536(7616):285–91. 10.1038/nature19057 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref020] 20.Gao GF, Parker JS, Reynolds SM, Silva TC, Wang LB, Zhou W, et al. Before and After: Comparison of Legacy and Harmonized TCGA Genomic Data Commons' Data. Cell Syst. 2019;9(1):24–34 e10. 10.1016/j.cels.2019.06.006 . [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref021] 21.Liu S, Hou W, Sun T, Xu Y, Li P, Yue B, et al. Genome-wide mining and comparative analysis of microsatellites in three macaque species. Mol Genet Genomics. 2017;292(3):537–50. 10.1007/s00438-017-1289-1 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref022] 22.Merkel A, Gemmell N. Detecting short tandem repeats from genome data: opening the software black box. Brief Bioinform. 2008;9(5):355–66. 10.1093/bib/bbn028 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref023] 23.Pandey M, Kumar R, Srivastava P, Agarwal S, Srivastava S, Nagpure NS, et al. WGSSAT: A High-Throughput Computational Pipeline for Mining and Annotation of SSR Markers From Whole Genomes. J Hered. 2018;109(3):339–43. 10.1093/jhered/esx075 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref024] 24.Nie X, Huang C, You C, Li W, Zhao W, Shen C, et al. Genome-wide SSR-based association mapping for fiber quality in nation-wide upland cotton inbreed cultivars in China. BMC Genomics. 2016;17:352 10.1186/s12864-016-2662-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref025] 25.Raney BJ, Dreszer TR, Barber GP, Clawson H, Fujita PA, Wang T, et al. Track data hubs enable visualization of user-defined genome-wide annotations on the UCSC Genome Browser. Bioinformatics. 2014;30(7):1003–5. 10.1093/bioinformatics/btt637 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref026] 26.Willems T, Zielinski D, Yuan J, Gordon A, Gymrek M, Erlich Y. Genome-wide profiling of heritable and de novo STR variations. Nat Methods. 2017;14(6):590–2. 10.1038/nmeth.4267 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref027] 27.Highnam G, Franck C, Martin A, Stephens C, Puthige A, Mittelman D. Accurate human microsatellite genotypes from high-throughput resequencing data using informed error profiles. Nucleic Acids Res. 2013;41(1):e32 10.1093/nar/gks981 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref028] 28.Sheng Q, Yu H, Oyebamiji O, Wang J, Chen D, Ness S, et al. AnnoGen: Annotating Genomewide Pragmatic Features. Bioinformatics. 2020. 10.1093/bioinformatics/btaa027 . [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref029] 29.Lin W, Piskol R, Tan MH, Li JB. Comment on "Widespread RNA and DNA Sequence Differences in the Human Transcriptome". Science. 2012;335(6074). 10.1126/science.1210624 WOS:000301531600026. [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref030] 30.Schrider DR, Gout JF, Hahn MW. Very Few RNA and DNA Sequence Differences in the Human Transcriptome. Plos One. 2011;6(10). ARTN e25842 10.1371/journal.pone.0025842 WOS:000295976000044. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref031] 31.Lee JH, Ang JK, Xiao X. Analysis and design of RNA sequencing experiments for identifying RNA editing and other single-nucleotide variants. RNA. 2013;19(6):725–32. 10.1261/rna.037903.112 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref032] 32.Liu Z, Zhang J. Human C-to-U Coding RNA Editing Is Largely Nonadaptive. Mol Biol Evol. 2018;35(4):963–9. 10.1093/molbev/msy011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref033] 33.McLean CY, Bristor D, Hiller M, Clarke SL, Schaar BT, Lowe CB, et al. GREAT improves functional interpretation of cis-regulatory regions. Nature biotechnology. 2010;28(5):495–501. 10.1038/nbt.1630 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref034] 34.Otlu B, Firtina C, Keles S, Tastan O. GLANET: genomic loci annotation and enrichment tool. Bioinformatics. 2017;33(18):2818–28. 10.1093/bioinformatics/btx326 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref035] 35.Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 2012;40(Database issue):D930–4. 10.1093/nar/gkr917 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref036] 36.Picardi E, D'Erchia AM, Lo Giudice C, Pesole G. REDIportal: a comprehensive database of A-to-I RNA editing events in humans. Nucleic Acids Res. 2017;45(D1):D750–D7. 10.1093/nar/gkw767 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref037] 37.Kiran A, Baranov PV. DARNED: a DAtabase of RNa EDiting in humans. Bioinformatics. 2010;26(14):1772–6. 10.1093/bioinformatics/btq285 . [DOI] [PubMed] [Google Scholar]

[pcbi.1007968.ref038] 38.Li Y, Shi W, Wasserman WW. Genome-wide prediction of cis-regulatory regions using supervised deep learning methods. BMC Bioinformatics. 2018;19(1):202 10.1186/s12859-018-2187-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref039] 39.Morgulis A, Coulouris G, Raytselis Y, Madden TL, Agarwala R, Schaffer AA. Database indexing for production MegaBLAST searches. Bioinformatics. 2008;24(16):1757–64. 10.1093/bioinformatics/btn322 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref040] 40.Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38(16):e164 10.1093/nar/gkq603 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref041] 41.Liu GE, Weirauch MT, Van Tassell CP, Li RW, Sonstegard TS, Matukumalli LK, et al. Identification of conserved regulatory elements in mammalian promoter regions: a case study using the PCK1 promoter. Genomics Proteomics Bioinformatics. 2008;6(3–4):129–43. 10.1016/S1672-0229(09)60001-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref042] 42.Penzkofer T, Jager M, Figlerowicz M, Badge R, Mundlos S, Robinson PN, et al. L1Base 2: more retrotransposition-active LINE-1s, more mammalian genomes. Nucleic Acids Res. 2017;45(D1):D68–D73. 10.1093/nar/gkw925 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref043] 43.Mir AA, Philippe C, Cristofari G. euL1db: the European database of L1HS retrotransposon insertions in humans. Nucleic Acids Res. 2015;43(Database issue):D43–7. 10.1093/nar/gku1043 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref044] 44.Yu Q, Zhang W, Zhang X, Zeng Y, Wang Y, Wang Y, et al. Population-wide sampling of retrotransposon insertion polymorphisms using deep sequencing and efficient detection. Gigascience. 2017;6(9):1–11. 10.1093/gigascience/gix066 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pcbi.1007968.ref045] 45.Morris JA, Gardner MJ. Calculating confidence intervals for relative risks (odds ratios) and standardised ratios and rates. Statistics in Medicine. 1988;296:1313–6. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Non-canonical RNA-DNA differences and other human genomic features are enriched within very short tandem repeats

Hui Yu

Shilin Zhao

Scott Ness

Huining Kang

Quanhu Sheng

David C Samuels

Olufunmilola Oyebamiji

Ying-yong Zhao

Yan Guo

Roles

Abstract

Author summary

Introduction

Results

A census of MNR/DNR/TNR polytracts in human genome

Table 1. Summary statistics of three clades of polytracts in GRCh38.

Fig 1. Landscape of MNR/DNR/TNR polytracts in the human reference genome GRCh38.

Corroborated and extended biological relevance of DNR and TNR

Fig 2. Enhancers are significantly enriched with polytracts.

Fig 3. Length distributions of Trinucleotide repeats (TNR) in the human genome.

Non-canonical RNA-editing events are enriched within MNR adjacency or hinge sites

Table 2. RNA editing event classes.

Fig 4. Non-canonical RNA-DNA differences are enriched within polytracts and, more markedly, break-points (hinges).

Landscape of genomic feature locational enrichment within polytracts

Fig 5. Landscape of genomic feature enrichment in MNR, DNR, and TNR polytracts.

A software package to evaluate genomic feature enrichment within polytracts

Fig 6. Schema of software package Polytrap.

Table 3. Polytrap computational cost on example sessions.

Discussion

Conclusions

Materials and methods

Identification and categorization of polytracts

Locational enrichment analysis of polytracts with respect to localizable genomic features

Genomic features enrolled in the locational enrichment analysis

Miscellaneous statistical analyses

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Lilia M Iakoucheva

William Stafford Noble

Roles

Author response to Decision Letter 0

Decision Letter 1

Lilia M Iakoucheva

William Stafford Noble

Roles

Author response to Decision Letter 1

Decision Letter 2

Lilia M Iakoucheva

William Stafford Noble

Roles

Acceptance letter

Lilia M Iakoucheva

William Stafford Noble

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases