Skip to main content
NCBI home page
NAR Genomics and Bioinformatics logoLink to NAR Genomics and Bioinformatics
. 2026 Feb 16;8(1):lqag017. doi: 10.1093/nargab/lqag017

FragmentFinder—a user-friendly, Windows-based tool for identifying and characterizing short RNAs excised from any noncoding RNA

Jeffrey D DeMeis 1, Courtlyn M Breaux 2, Allison H Powell 3, Nick J Campbell 4, Dominika Houserova 5, Jingshan Huang 6,7, Glen M Borchert 8,✉
PMCID: PMC12907730  PMID: 41704564

Abstract

Noncoding RNAs <200 nucleotides (nt) in length are referred to as short noncoding RNAs (sncRNAs) and include microRNAs (miRNAs), piwi-interacting RNAs, small nucleolar RNAs, transfer RNAs, etc. One striking example of the regulatory capabilities of sncRNAs comes from a group of small yet potent RNAs called miRNAs. MiRNAs are ∼20-nt RNAs excised from longer pre-miRNA hairpins, and to date, thousands of miRNAs have been identified across an array of species with specific roles for miRNAs defined in virtually every cellular activity (e.g. growth, differentiation, apoptosis, and disease). Importantly, studies aimed at evaluating the transcriptomic changes of miRNAs have now revealed the existence of miRNA-like fragments derived from other types of sncRNAs and suggest similar regulatory capacities may be associated with these novel sncRNA fragments. Unfortunately, many biologically relevant sncRNA-excised fragments remain uncharacterized due to their routine exclusion during initial miRNA characterizations as “sncRNA degradation products” as well as nearly all sncRNA informatic analyses continuing to solely assess annotated miRNA expressions. To address this, several platforms aimed at identifying novel sncRNA fragments have recently been developed. That said, the principal analytical tools currently employed to characterize novel sncRNA fragments often require significant computational expertise hindering their widespread utilization. As such, the development of a user-friendly platform, requiring minimal programming experience yet capable of identifying and characterizing RNA fragments excised from any sncRNA from any species is highly desirable and potentially impactful. In light of this, we have developed FragmentFinder—an intuitive, Windows-executable resource designed to require absolutely no computational background and capable of accurately characterizing all (annotated and unknown) sncRNA-derived RNAs within a raw small RNA sequencing file in real time.

Introduction

The term noncoding RNA (ncRNA) encompasses a wide variety of RNA species and can be stratified into two subgroups based on transcript length: (i) long noncoding RNAs (lncRNAs) consisting of RNAs >200 nucleotides (nt) in length and (ii) short noncoding RNAs (sncRNAs) comprising transcripts smaller than 200 nt in length such as small nucleolar RNAs (snoRNAs), vault RNAs (vtRNAs), and transfer RNAs (tRNAs) [1]. In addition to lncRNAs and traditional sncRNAs, rapid technological advancements in transcriptomics, particularly the development of deep sequencing techniques, have now also revealed the existence of a wide range of sncRNA-derived RNA fragments similar to microRNAs (miRNAs) consisting of ∼20-nt stretches of RNA specifically excised from longer parental transcripts [e.g. tRNA fragments (tRFs), snoRNA-derived RNAs (sdRNAs), short vault RNAs (svtRNAs), rRNA fragments (rRFs), short Y RNA-excised RNAs (ysRNAs), etc.] [2, 3, 4]. Importantly, recent studies strongly suggest that many of these RNA fragments exert functions wholly distinct from their longer precursor sncRNAs [3–6]. That said, many of these miRNA-sized RNAs remain undescribed due to their routine exclusion during initial miRNA characterizations as “sncRNA degradation products” and as their expressions often prove highly conditional (i.e. environmental stress, disease, tissue type, etc.) [7]. Thus, the development of a high-throughput platform that is (i) capable of identifying and characterizing short RNAs excised from any sncRNA from any species and (ii) designed for biologists possessing little to no programming experience is highly desirable and potentially impactful. As such, the work detailed herein addresses this need by describing FragmentFinder (FF)—an intuitive, Windows-executable resource for biologists capable of accurately characterizing all (both annotated and unknown) sncRNA-derived RNAs regardless of species.

FF overview

  • Identifies short RNAs specifically excised from longer sncRNAs based on positional enrichment of aligning reads.

  • Allows the user to upload their own database of potential precursor RNAs (e.g. all human sncRNAs <200 nt, all mouse snoRNAs, all Arabidopsis tRNAs, etc.).

  • Allows the user to upload any raw sncRNA/miRNA-sequencing (sncRNA/miRNA-seq) next-generation sequencing (NGS) output file. No processing or trimming of reads is required.

  • Provides direct visualization of sncRNA excisions with normalized expression in reads per million (RPM).

  • Results from separate FF analyses can be merged into one excel file using the TSV-Merger companion tool for direct comparison of all sncRNA fragment expressions across distinct sequencing files.

FF input

Upon opening the FF graphical user interface (GUI) the user will be prompted to specify an sncRNA-seq or miRNA-seq read file (.txt, .fasta, or .fastq extensions) (either uncompressed or zipped) followed by a FASTA-formatted database file (.fasta or .txt extension) containing user specified precursor transcripts of interest (Figs 1 and 2A and B). Major advantages of FF are that (i) it can handle raw RNA-seq reads without any user-dependent preprocessing regardless of library preparation specifics, and (ii) no specific size limitations with respect to input files are specified [although large read files (e.g. >10 Gb) will require significantly longer run times].

Figure 1.

Figure 1.

Open in a new tab

FF workflow. Sequence input (top). The user is prompted to import an sncRNA/miRNA-seq file and then a database file containing user specified precursor transcripts of interest. RNA fragment detection and visualization (middle). The user initiates preprocessing to identify NGS read-database pairs by clicking “Preprocess” followed by “Let’s Go!” to activate FF’s core alignment logic. Upon completion, fragments are detected by positional enrichment relative to their corresponding precursor transcripts. Data merging for high throughput analysis (bottom). After completing multiple runs of FF against a particular database, the user can employ the TSV-Merger companion tool to compare the expressions of specific sncRNA fragments across multiple runs.

Figure 2.

Figure 2.

Open in a new tab

Navigating FF GUI. (A) and (B) Image of FF GUI display prompting user to provide the input sncRNA/miRNA-seq and sncRNA database files, respectively. *, 16 GB of RAM is the recommended baseline for general use and basic productivity for users working with ≤10 GB sequencing files; 32 GB of RAM or more is recommended for analysis of >10 GB files. (C) Image of selectable expression vector (EV) table in FF GUI following run completion (see “FF implementation” under “Materials and methods” section for a description of EV determination). The user can select an sncRNA of interest and click “Plot Selected ncRNA” to produce an individual coverage plot. Stepwise instructions for using the FF GUI are listed to the right of their corresponding display prompts. Bullet points elaborate on input file requirements and how to locate and extract data from GUI.

FF output

After importing the sncRNA/miRNA-seq and database files, the user initiates preprocessing to identify NGS read-database pairs by clicking “Preprocess” followed by “Let’s Go!” to activate FF’s core alignment logic. Upon completion, FF will produce a .tsv output file named according to the input files in the same directory as the application. This file contains the metadata for all of the RNA fragments identified, including (i) the precursor transcripts from which the fragments are derived from along with their start–stop positions, (ii) their sequences (peak data), and (iii) their expressions in RPM. Importantly, any output .tsv file can be merged with any other FF output files run against the same precursor database to compare individual fragment expressions. More detailed instructions regarding the use of TSV-Merger are described in Supplementary File S1. In addition, the user can select any precursor transcript of interest followed by “Plot Selected ncRNA” to produce a graphical plot of all reads aligning to that transcript directly in the FF program, or alternatively, the user can export all graphical plots as a .pdf file by clicking “Export Plots” (Fig. 2C; see Fig. 3A for example plot).

Figure 3.

Figure 3.

Open in a new tab

FF identifies validated mature miRNA sequences. (A) Screenshot of FF GUI plot for hsa-mir-135b expression produced by using SRR8479191 as the input sncRNA/miRNA-seq file (as shown in Fig. 2). (B) Select example data taken from output .tsv file produced using SRR8479191 as the input sncRNA/miRNA-seq file. Output includes: database entry name (miRNA ID), normalized expression (RPM), Peak start, and Peak end, fragment sequence (Peak Data), and the total number of reads present in the sncRNA/miRNA-seq file (Total Reads in NGS file). Data corresponding to hsa-mir-135b-5p are denoted by red font. (C) Screenshot of miRBase [10] histogram for hsa-mir-135b (MI0000810) above the corresponding precursor transcript with miRBase-annotated mature miRNA (hsa-mir-135b-5p) sequence denoted by blue font [10] (middle) aligned to the peak sequence called by FF in our analyses (bottom).

Importantly, besides outlining FF’s installation and execution, the study presented herein also demonstrates FF’s utility via performing and detailing a thorough examination of a publically available miRNA-seq dataset initially generated to identify human miRNAs misexpressed in pancreatic cancer as compared to normal pancreatic tissue. Of note, in addition to independently identifying and confirming the misexpressions of specific miRNAs previously reported as being misexpressed in these pancreatic cancer datasets [8], FF also identifies a number of additional miRNA-sized fragments—excised from several other types of ncRNAs [e.g. ribosomal RNAs (rRNAs), tRNAs, and snoRNAs]—that are similarly misexpressed in pancreatic tumors relative to patient-matched controls.

Materials and methods

sncRNA/miRNA-seq analysis of pancreatic cancer patient tissues

FASTA-formatted sncRNA/miRNA-seq datasets were downloaded from the sequence read archive (SRA) database using the sra-toolkit. The SRA files used were derived from seven pancreatic cancer patient tumors (SRR8479189, SRR8479191, SRR8479193, SRR8479195, SRR8479197, SRR8479199, and SRR8479201) and patient-matched normal adjacent tissues (SRR8479188, SRR8479190, SRR8479192, SRR8479194, SRR8479196, SRR8479198, and SRR8479200) [8]. Each sequencing file was run through the FF application against a FASTA-formatted database consisting of (i) all sncRNAs annotated in Ensembl and (ii) human 12s, 16s, 18s, and 28s rRNA sequences divided into 105 nt pieces with 5 nt overlaps (Supplementary File S3). Notably resulting output .tsv files (Supplementary Files S4–S17) were merged using the companion tool TSV-Merger.exe. Merged data were then sorted by tissue type (control or tumor) and average fragment expression determined with and without pancreatic cancer (Supplementary File S2). The significance of differentially expressed fragments was determined by unpaired two-tailed t-test assuming equal variances.

FF implementation

FF was designed to specifically identify the longest perfect sequence match between each sequencing read in an sncRNA/miRNA-seq dataset and any of the potential ncRNA precursors found within a user-specified ncRNA database. To achieve this, FF first identifies all NGS reads which possess a perfect 18-nt match to any given database entry, then these matched reads (stored as temporary “matches.txt”) are subjected to a more thorough alignment to the database in which perfect alignments are extended up to 38 nt or until a mismatch is encountered. Qualified alignments corresponding to the longest perfect, uninterrupted commonalities between a database entry and a sequencing read are then used to build an EV for each database entry where an EV comprises an array of positional counts (i.e. the number of aligned reads mapping to a given position on the database entry). These EVs are then used to impute a coverage signal for the corresponding database entry from which “peaks” are called. To be recognized as a peak region coverage was required to exhibit a prominence ≥85% of the standard deviation for the entire database entry. To identify a peak’s full length, regions which meet the prominence threshold are expanded until there is a detected drop in coverage, i.e. a peak end is called when a drop of 10% of the total coverage signal variability is encountered. Once refined peak positions are obtained, the normalized expression (RPM) of each peak is computed and all peaks of at least 5 RPM are appended into an output .tsv file with each row containing data corresponding to an individual peak, including the mapped database entry, RPM, peak start position, peak stop position, and the peak nucleotide sequence. Database ncRNAs not found to contain peaks based on the preceding criterion are not included in output .tsv files. Additionally, we have incorporated two distinct statistical measures to (i) infer the likelihood of occurrence for a specified peak relative to the null distribution of that FF run, using a permutation approach consisting of one million permutations (i.e. number of times data are resampled) and (ii) rank peaks by z-score which is calculated for each individual peak by subtracting precursor average baseline expression (excluding peak expression and 5 nt flanks) from peak maximum expression and then dividing by the standard deviation of precursor baseline.

FF statistical analyses

For the permutation approach, to estimate a baseline (null) distribution, the region around each qualified peak (as defined under “FF implementation” section)± 5 nt is removed from every precursor to create a clean background, and all such background values are pooled to form a global baseline expression set. An empirical permutation process is then used to compute P-values: for each peak value, N (1000 000) samples are randomly drawn with replacement from the global background, and the number of sampled values greater than or equal to the observed peak is counted. The empirical P-value is finally calculated using a continuity-corrected permutation test formula pi = (1 + C)/(1 + N), which prevents P = 0 and yields stable estimates even when the observed peak exceeds all sampled background values. For the z-score approach, each qualified peak is evaluated by calculating a z-score relative to the baseline of its corresponding precursor. The z-score is computed as z = (xpeak − baselinemean)/baselinestandard deviation, where xpeak is the read count at the center of the peak, baselinemean is the average of all baseline expression values excluding all peaks and ± 5 nt, and baselinestandard deviation is the standard deviation of those same baseline values. After computing z-scores for all peaks across a dataset, values are collected and sorted, and each peak’s empirical P-value is calculated using the expression P = (1 + number of z-scores greater than or equal to zi)/(1 + total number of peaks). This generates a right-tailed empirical P-value that reflects how likely it is to observe a z-score as large as the one measured under the empirical distribution, and the “plus one” added to both the numerator and denominator serves as a continuity correction to prevent P = 0.

FF database construction

After specifying an miRNA-seq read file, FF will ask the user to provide a database file containing user specified precursor transcripts of interest (Fig. 2). FF will accept database files with either a .fasta or .txt file extension. Regardless of file extension, individual database sequences must be entered in standard FASTA format with each entry consisting of two lines. Line 1 (the header) begins with a “>” followed by all desired information corresponding to the entry (e.g. name, genomic location, etc). There is no limit on header length and all characters occurring between the initial “>” and the first hard return will be taken as the ID for a given entry. Line 2 should contain nothing other than a linear string of characters consisting of the nucleotide sequence (5′ to 3′) corresponding to the transcript identified on the preceding line (see e.g. Fig. 2B). All characters between the hard return at the end of Line 1 and the hard return at the end of Line 2 will be taken as the specified transcript sequence. Importantly, database sequences should only contain, A, G, T, and C characters. FF does not recognize IUPAC ambiguity codes. As such, databases should only contain specific sequences, not consensus sequences with variable nucleotides. In addition, individual database sequences should not exceed 300 nt in length to minimize bias arising from the requirement for a peak to exhibit a prominence ≥85% of the standard deviation for the entire database entry sequence. If the inclusion of sequences longer than 300 nt is desired (e.g. rRNAs), these should be split into shorter, overlapping pieces. As an example, the database utilized in this study (Supplementary File S3) contains human 12s, 16s, 18s, and 28s rRNA sequences divided into 105 nt pieces with 5 nt overlaps. A length of 105 nt was selected as this corresponds to the average database sequence length, and 5 nt overlaps were included to minimize the chances of missing a fragment excised from locations where longer sequences were split. Finally of note, while we find most human noncoding RNA annotations are fairly well defined, annotated transcript positions are often less accurate in less well-characterized species. We find this can result in artificial 5′ and/or 3′ ncRNA sequence truncations causing FF to miss sncRNAs excised from ncRNA ends. As such, we suggest including 20 nt genomic flanks to database ncRNA sequences whenever possible.

FF data merging tool

After running multiple files with FF, the user can utilize TSV-Merger.exe to group peaks and accompanying metadata together. When a run is initiated, the user will be prompted to specify the directory containing the FF output .tsv files followed by the file location of the database .txt file used during alignment. All .tsv files are joined into a single data frame and sorted by “miRNA ID” and “peak start” and are temporarily saved as t1.tsv. From t1.tsv, nearby peaks are grouped, and a new group is initiated whenever the miRNA ID changes, or the ensuing start is >10 nt of the prior start. The reference sequences are then loaded as a dictionary {database ID: sequence} and finally consensus peaks are determined for each group finding the minimum start and maximum end positions, the nucleotide sequence corresponding to this peak position is then extracted from the reference database and builds a list of RPMs for this peak across all files being merged returning one row for the group consisting of miRNA ID | Peak Start | Peak End | Files : RPM| Peak Data. Where “files: RPM” consists of a list of dictionaries corresponding to the RPM for that group in each input file and these resulting RPMs obtained from each NGS read file are deposited in their own columns in the final merged output .tsv. Additionally, the merge tool automatically performs a quality check where the final merged peak sequences are cross referenced back to the database to identify any alternative regions of origin for each peak sequence. Results are output into a “checked.tsv” file which contains all of the merged data in addition to columns concerning the number of possible alternate origin positions (Alt_Origins_Count) and a column listing each possible origin for the peak sequence in that row.

FF executable configuration

FF.exe and TSV-Merger.exe are compatible with all currently supported Windows and Linux operating systems and require no additional software installations. FF was developed into a Windows-compatible executable package using the following technology stack: Programming languages: python 3.13 and cython 3.1.0. Other tools and supporting technologies: Visual Studio Build Tools, PyQt5, matplotlib, numpy, scipy, ahocorasick, Biopython, pandas, Glob, Ast, and Re. Hardware: Intel® Core™ i7-10700 CPU @ 2.90 GHz, 2904 Mhz, 128 GB RAM, and 1 TB local fixed disk.

Results

FF identifies validated mature miRNA sequences

To confirm that FF can successfully identify biologically relevant fragments, we used FF to analyze 14 sncRNA/miRNA-seq files produced by a previous study aimed at identifying miRNAs highly expressed in pancreatic tumors relative to patient-matched control tissues (SRP181693) [8]. We selected this study primarily due to these files also being independently analyzed in another subsequent study employing “tsRNAsearch”, a nextflow pipeline for identifying tRNA and ncRNA fragments from small RNA-seq files [9]. We found the miRNAs FF identified as being significantly enriched in pancreatic tumors relative to patient-matched control tissues well agreed with those identified in the initial study aimed at identifying miRNAs misexpressed in pancreatic cancer as compared to normal pancreatic tissue. As a specific example, the initial study by Yang et al. identified mir-135b-5p to be significantly upregulated in, and a direct mechanistic contributor to, pancreatic cancer, and in agreement with this study, FF analysis of these same datasets also identified hsa-mir-135b-5p as being specifically excised from the hsa-mir-135b hairpin in these samples [10] (Fig. 3). What is more, after running FF on each of the 14 sncRNA/miRNA-seq files generated by the original study, we used the TSV-Merger companion tool (Fig. 4A) to compare the expressions of all sncRNA fragments identified in these runs, and similarly found hsa-mir-135b-5p to be expressed >8× more highly in pancreatic tumors than in controls (P = .047). Of note, FF also identified 70 other miRNAs significantly differentially expressed in pancreatic tumors relative to patient-matched control tissues (P < .05) (Supplementary File S2). We selected 10 of these that were appreciably expressed (i.e. ≥25 RPM in either control or tumor tissues) for further analysis, and notably, find that each of these have been previously reported as being significantly differentially expressed in either pancreatic cancer patient tissues or pancreatic cancer cell lines [8, 9, 11–18, 19] (Fig. 4B). Finally, as an additional means of confirming FF accuracy, we elected to align the sequences called by FF as being excised from these miRNA precursors (Fig. 4B) to the reported, experimentally verified mature miRNA sequences annotated in miRBase [10]. Importantly, we find FF identified miRNA sequences match miRBase-annotated mature miRNAs with high-fidelity (>95% identity) (Fig. 4C).

Figure 4.

Figure 4.

Open in a new tab

TSV-Merger companion tool enables high throughput comparison of expression data produced by FF. (A) Screenshot of TSV-Merger GUI with buttons prompting the user to specify the directory housing output .tsv files produced by database-matched runs of FF. (B) Differentially expressed miRNAs averaging ≥25 RPM (either in normal or tumor tissue) identified by TSV-Merger compared across the 14 sncRNA/miRNA-seq files generated by a study aimed at identifying miRNAs highly expressed in pancreatic tumors relative to patient-matched control tissues (SRP181693). Sequencing files were downloaded from the SRA database (SRP181693) and run through FF against a database containing transcripts of all Ensembl-annotated human sncRNAs. Excised fragment sequences are included under “Peak Data,” and normalized miRNA expressions (RPM) are listed in columns titled by the respective SRA input file name. pvalue: Statistical significance of differentially expressed miRNA determined using an unpaired two-tailed t-test assuming equal variances. PMID: PubMed IDs of prior studies reporting an miRNA as significantly differentially expressed in either pancreatic cancer patient tissues or pancreatic cancer cell lines. Data corresponding to hsa-mir-135b-5p are denoted by red font. (C) Alignments between miRBase-annotated mature miRNAs and differentially expressed miRNAs associated with pancreatic cancer. Image of alignments between miRBase-annotated mature miRNA sequences (top) and RNA-derived fragments identified by FF (bottom). Alignment corresponding to hsa-mir-135b-5p is denoted by red font.

FF identifies previously reported miRNA-sized fragments excised from other types of sncRNAs

After confirming FF could reliably identify verified mature miRNA sequences in sncRNA/miRNA-seq datasets, we next examined FF’s ability to similarly identify miRNA-sized fragments excised from a variety of other types of sncRNAs such as rRNAs, tRNAs, snoRNAs, etc. Analysis of the same merged .tsv file used to identify the miRNAs depicted in Fig. 4 revealed 25 rRFs, 1 tRF, 1 svtRNA, and 13 sdRNAs similarly differentially expressed between patient tumors and controls (P < .05) and averaging ≥25 RPM (either in normal or tumor tissue). Importantly, we find that each of the 25 significantly differentially expressed rRFs identified by FF have previously been reported and are annotated in the primary ribosomal RNA fragment database, MINRbase [20] (Fig. 5A). Additionally, although only one tRF was identified as being significantly differentially expressed between patient tumors and controls, 176 unique tRFs were identified that were appreciably expressed in multiple samples, and we find the 114 of these have also previously been reported and are likewise annotated in the primary tRF database, MINTbase [20, 21] (Fig. 5B). Moreover, while major comprehensive databases are not currently available for RNA fragments excised from most other types of sncRNAs, we find several reports independently corroborating a number of the ysRNAs, svtRNAs, and sdRNAs identified by FF [22–24] in these samples (Fig. 5C). Finally, just as with miRNA calls, as an additional means of supporting FF’s accuracy, we aligned the sequences called by FF to sequences annotated in corresponding databases or contained in published reports (Fig. 6).

Figure 5.

Figure 5.

Open in a new tab

FF identifies previously reported sncRNA-derived RNAs excised from a variety of precursors. Sequencing files were downloaded from the SRA database (SRP181693) and run through FF against a database containing transcripts of all Ensembl-annotated sncRNA. Output .tsv files were merged using TSV-Merger companion tool. (A) Merged expression data of select rRFs identified by FF and independently reported in MINRbase rRF database [20]. (B) Merged expression data of select tRFs identified by FF and independently reported in the MINTbase tRF database [21]. (C) Merged expression data of RNA fragments derived from other types of sncRNA precursors such as yRNA (top), vtRNA (middle), and snoRNA (bottom). Pubmed IDs corresponding to studies which have previously reported these RNA fragments [4, 22, 24]. Numerical data listed in columns titled by SRR file number represent the normalized expression (RPM) of each fragment within that sequencing file and accompanying fragment sequences are listed under “Peak Data.”

Figure 6.

Figure 6.

Open in a new tab

Alignments between reported sncRNA-derived fragments and excision products identified by FF. Alignments between previously reported sncRNA-derived fragments (top) and fragments identified by FF (bottom). rRF and tRF sequences were obtained from MINRbase and MINTbase, respectively [20, 21]. ysRNA, svtRNA, and sdRNA sequences were obtained from previously published studies [22, 23, 24].

FF comparison to existing tools

To evaluate the utility of FF, we conducted a comparative analysis involving 11 existing sncRNA characterization platforms (Table 1). Given that FF was designed specifically to be a user-friendly tool for characterizing the widest possible range of sncRNA fragments across species, our assessment examined whether a given program is (i) user-friendly (i.e. accessed as an executable package or requires minimal command-line interface and results are easily interpreted) and (ii) able to identify different types of sncRNA-derived fragments from different organisms. Most notably, we find three tools (Oasis2.0, CPSS2.0, and SPAR) which we consider to be similarly “user-friendly” but find that each of these were designed to characterize full-length sncRNAs rather than sncRNA-derived fragments. Conversely, we find three programs (sRNAfrag, sRNAtools, and FlaiMapper) largely capable of comparably identifying sncRNA fragments from any precursor sncRNA much like FF, and another tool (SPORTS1.0) capable of identifying fragments excised from tRNAs and rRNAs, but find that each of these programs requires direct command-line execution/considerable computational expertise. Of note, we directly compared FF and sRNAfrag run times as sRNA frag is the only other currently available program capable of identifying and comparing the expressions of novel sncRNAs excised from any ncRNA precursor from any species. Strikingly, we found Windows-executable FF run times for the 14 SRA files examined in this study averaged 0.17 sRNAfrag run times for these same files run on identical hardware (detailed in “FF executable implementation” under “Materials and methods” section) and speculate this is likely due to FF requiring perfect alignments whereas sRNAfrag allows mismatches. What is more, we find FF the only currently available program (i) which does not require the user to specify NGS library oligo/adapter sequences and (ii) providing both a user-friendly interface and the capacity to identify sncRNA fragments processed from any sncRNA precursor/species.

Table 1.

Comparison of FragmentFinder to existing sncRNA analysis platforms

Program Input User-friendly Library oligo seqs required? Known full-length sncRNA expressions Novel full-length sncRNA discovery Novel sncRNA fragment discovery Differential expression species
FragmentFinder Raw Yes No No No Yes Yes User
sRNAfrag Raw No Yes No No Yes Yes User
sRNA toolbox Raw No Yes Yes Yes No sRNAde 175
Oasis2.0 Raw Yes Yes Yes Yes No Yes 14
sRNAtools Raw No Yes Yes Yes Yes Yes 21
CPSS2.0 Pre-processed Yes Yes Yes Yes No Yes 48
SPAR Pre-processed Yes Yes Yes Yes No Yes 2
sRNAnalyzer Raw No Yes Yes No No No User
SPORTS1.0 Raw No Yes Yes User tRFand rRF only Yes 68
DEUS Raw No Yes Yes Yes User Yes User
FlaiMapper Pre-processed No Yes No No Yes No User
featureCounts Pre-processed No Yes User No User No User
Open in a new tab

Various features offered by FF were compared to other existing tools, including sRNAfrag [25], sRNA toolbox [26], Oasis2.0 [27], sRNAtools [28], CPSS2.0 [29], SPAR [30], sRNAnalyzer [31], SPORTS1.0 [32], DEUS [33], FlaiMapper [34], and featureCounts [35]. Features examined were as follows: input: form of input RNA-seq dataset—either raw (direct NGS output) or pre-processed (e.g. requires BAM file or adapter trimming); user-friendly: if interactive and user-friendly results are generated directly; library oligo sequences req: if user knowledge of NGS oligo sequences is required; known full-length sncRNA expressions: detection and quantification of known sncRNAs; novel full-length sncRNA expressions: detection and quantification of novel sncRNAs; novel sncRNA fragment discovery: detection and quantification of novel sncRNA fragments; differential expression: ability of the tool to integrate expression data from multiple files (“sRNAde” denotes that expression analyses can be performed in parallel); and species: number of species available for analysis. “User” denotes that the tool has the capacity to perform a given task regardless of species, however, requires user input or user-directed change to program’s code and/or advanced settings.

Discussion

It is well established that sncRNAs such as snoRNAs, tRNAs, 5s rRNAs, etc., possess biologically relevant roles [36, 37, 38] and it is becoming increasingly clear that many of these sncRNAs possess subsequences which are consistently and specifically excised from their precursors and serve cellular roles distinct from their parental transcripts [3, 5, 6, 39]. For instance, recent studies have identified that miRNA-like excision products derived from snoRNAs, termed “sdRNAs,” can influence cancer phenotypes such as cellular invasion and drug sensitivity through regulation of select target genes [6, 40]. That said, whereas the cost of sncRNA/miRNA-seq is now reasonable for most active research programs, programs necessary for the interpretation of these sequencing datasets typically require significant computational expertise and resources markedly hindering widespread utilization of these tools. As new high-throughput small RNA sequencing strategies [41] continue to make small RNA-seq faster and less expensive, there is a clear need for biologist-accessible tools capable of analyzing large amounts of small RNA-seq data to detect and characterize all small RNA genes, including small RNA excision products. As such, the necessity for development of real-time, user-friendly platforms capable of making the identification and characterization of the sncRNAome accessible to biologists lacking significant computational expertise is strikingly apparent. To address this need, we have developed FF a user-friendly, Windows-executable platform for efficient and accurate detection and quantification of sncRNA-derived RNAs from raw sncRNA-seq files and requiring absolutely no dataset preprocessing or knowledge of library adapters/oligonucleotides. In short, FF represents a powerful new resource, designed specifically for users with no computational background, to enable widespread identification and characterization of sncRNA-derived RNAs.

Importantly, in addition to outlining FF’s installation and execution, this work presented in this study also demonstrated FF’s utility and accuracy via performing a thorough examination of an miRNA-seq dataset initially generated to identify human miRNAs misexpressed in pancreatic cancer. In all, FF identified 71 miRNAs significantly differentially expressed in pancreatic tumors relative to patient-matched control tissues (P < .05) (Supplementary File S2). We find that the majority of mature miRNA sequences FF identified as being specifically excised from human miRNA precursors (i) match experimentally verified, miRBase-annotated mature miRNA sequences nearly perfectly (>95% identity) (Fig. 4C) and (ii) have been previously reported as being significantly differentially expressed in either pancreatic cancer patient tissues or pancreatic cancer cell lines [8, 9, 11–18, 19] (Fig. 4B). That said, while accurately replicating previously reported mature miRNA expressions in these datasets is an important confirmation of FF’s accuracy, FF’s actual utility is better evidenced through its identification of a number of additional miRNA-sized fragments—excised from several other types of ncRNAs (e.g. rRNAs, tRNAs, and snoRNAs)—that are similarly misexpressed in pancreatic tumors relative to patient-matched controls. In addition to identifying differentially expressed miRNAs (Fig. 4), FF also identified 25 rRFs, 1 tRF, 1 svtRNA, and 13 sdRNAs similarly differentially expressed between patient tumors and controls (P < .05). Importantly, we find that each of the 25 rRFs have previously been reported and are annotated in the primary ribosomal RNA fragment database, MINRbase [20] (Fig. 5A). Additionally, FF largely agreed with a 2021 analysis performed by Donovan et al. [9] examining the same datasets used in this study. Similar to the findings reported by Donovan et al., FF identified 176 unique tRFs appreciably expressed in multiple samples but only one of these to be significantly differentially expressed between patient tumors and controls. What is more, we find that 114 of these tRFs identically match tRFs annotated in the primary tRF database, MINTbase [20, 21] (Fig. 5B). Finally, we find several of the ysRNAs, svtRNAs, and sdRNAs identified by FF [22, 23, 24] in these samples (Fig. 5C) perfectly correspond to sequences annotated in corresponding databases or contained in published reports (Fig. 6).

Notably, pancreatic acinar cells, which constitute ∼90% of pancreatic tissue mass and some pancreatic cancers, express high levels of RNases resulting in significant degradations frequently being observed in RNAs isolated from these tissues. That said, the aforementioned corroboration of the excised sequences identified by FF with those published in previous reports strongly indicates that these sequences represent genuine, specific excisions and not random degradation products. As such, the results presented in this study strongly suggest FF’s requirement for an excised sequence to exhibit a prominence ≥85% the standard deviation of the full-length ncRNA sequence can efficiently differentiate between specific excisions and random degradation products.

Limitations

(i) Perhaps the most significant limitation of our program and many others is in the handling of multimapping or ambiguous reads. That is, an inability to determine the true source of a sequencing read possessing two or more perfect matches to a reference database. We find that FF arbitrarily attributes ambiguous sequencing reads to the last instance of a perfect match in the database, therefore swapping the order of redundant database entries will incidentally alter coverage plots and expression counts for these select transcripts. To address this limitation, we incorporated a quality check within the merge companion tool whereby peak sequences listed in the merged .tsv file are cross referenced with the input database file to identify any potential alternative origins for a given fragment and list these in the merge output. Notably, however, we suggest this limitation represents a fairly minor concern as we find that only 1 of the 71 miRNAs we identified as significantly differentially expressed in pancreatic tumors relative to patient-matched control tissues (Supplementary File S2) were potentially excised from multiple precursors excluding related hairpins (e.g. mir-103a-1 and mir-103a-2). (ii) While FF does provide statistical measures to elucidate significance between distinct fragments called from a single run (permutation P-value and z-score rank), it does not provide a statistical measure regarding differential expression of an individual fragment across merged output files. Thus, the user must conduct their own statistical analyses. (iii) Additionally, the FF GUI does not enable batch runs, requiring the user to close and reopen the application with each subsequent run; however, we perceive this to be a minor limitation given FF’s run speed. Finally, (iv) we also unexpectedly encountered a file limit for the TSV-Merger companion tool. While we have been able to successfully merge up to 85 separate files produced by FF (data not shown), we find TSV-Merger unable to successfully merge analyses comparing more than 85 files using the hardware detailed in “FF executable implementation” under “Materials and methods” section. While we assume this limit can be circumvented by employing more advanced or additional hardware, this has not been directly confirmed.

Advantages

Currently, most of the existing tools for sncRNA-derived fragment detection either (i) focus almost exclusively on miRNAs and/or only evaluate a limited predefined set of sncRNAs, and as such are not capable of fully defining small RNA-seq sncRNA fragment profiles and/or differences among datasets (e.g. miRDeep [42], miRSpring [43], miRanalyzer [44], sRNAnalyzer [31], Oasis2.0 [27], SPAR [30], etc.) or (ii) are capable of characterizing novel sncRNA fragments and their expressions but require extensive computational expertise for utilization (e.g. sRNAfrag [25], FlaiMapper [34], SPORTS [32], DEUS [33], etc.). Considering these limitations, we have designed FF to allow users to look for sncRNA excisions from any sncRNA precursors and to be readily executable with little to no computational expertise. In stark contrast to most existing platforms, FF identifies fragments excised from any collection of transcripts in a user-provided transcriptome uploaded as an sncRNA-seq file. In addition, FF’s merge companion tool enables direct comparison of individual fragment expressions across distinct datasets. Importantly, there are currently no user-friendly resources that offer comprehensive sncRNA fragment profiling and discovery comparable to FF. Although one platform, sRNAtools [28], does offer many of FF’s features, FF requires significantly less run time and distinguishes itself by providing more intuitive, versatile, and user-accessible results.

Summary

Importantly, although studies aimed at evaluating the transcriptomic changes of miRNAs have revealed the existence of an array of miRNA-sized fragments derived from several distinct types of ncRNAs, the vast majority of ncRNA-excised fragments remain uncharacterized primarily due to no user-friendly, widely applicable platforms capable of identifying novel, sncRNA fragments being currently available. As such we suggest FF represents a powerful new tool for identifying biologically relevant sncRNA fragments as FF enables users with no computational background to quickly and easily analyze, visualize, and compare small RNA-seq datasets in order to generate clear, informative results. With an interactive user-friendly interface, FF is the first Windows-executable resource that provides users the ability to upload unmodified NGS datasets and any database of potential precursor sncRNAs to perform comprehensive novel sncRNA fragment identifications and expression analyses in real time.

Supplementary Material

lqag017_Supplemental_Files
lqag017_supplemental_files.zip (1.6MB, zip)

Acknowledgements

We thank the University of South Alabama College of Medicine for ongoing support.

Jeffrey D. DeMeis (Conceptualization [equal], Data curation [lead], Formal analysis [lead], Methodology [equal], Software [equal], Validation [equal], Writing – original draft [lead], Writing – review & editing [equal]), Courtlyn M. Breaux (Software [supporting], Writing – review & editing [supporting]), Allison H. Powell (Software [supporting], Writing – review & editing [supporting]), Nick J. Campbell (Software [equal], Writing – review & editing [supporting]), Dominika Houserova (Conceptualization [equal], Software [equal], Writing – review & editing [equal]), Jingshan Huang (Software [supporting], Writing – original draft [supporting], Writing – review & editing [supporting]), and Glen M. Borchert (Conceptualization [lead], Data curation [equal], Formal analysis [equal], Funding acquisition [lead], Investigation [lead], Methodology [equal], Project administration [lead], Resources [lead], Software [supporting], Supervision [lead], Validation [equal], Visualization [lead], Writing – original draft [supporting], Writing – review & editing [equal])

Contributor Information

Jeffrey D DeMeis, Department of Pharmacology, Frederick P. Whiddon College of Medicine, University of South Alabama, Mobile, AL 36688, United States.

Courtlyn M Breaux, Computer Science, School of Computing, University of South Alabama, Mobile, AL 36688, United States.

Allison H Powell, Computer Science, School of Computing, University of South Alabama, Mobile, AL 36688, United States.

Nick J Campbell, Computer Science, School of Computing, University of South Alabama, Mobile, AL 36688, United States.

Dominika Houserova, Center for Cellular and Molecular Therapeutics at Children’s Hospital of Philadelphia, 3501 Civic Center Boulevard, Philadelphia, PA 19104, United States.

Jingshan Huang, Department of Pharmacology, Frederick P. Whiddon College of Medicine, University of South Alabama, Mobile, AL 36688, United States; Computer Science, School of Computing, University of South Alabama, Mobile, AL 36688, United States.

Glen M Borchert, Department of Pharmacology, Frederick P. Whiddon College of Medicine, University of South Alabama, Mobile, AL 36688, United States.

Supplementary data

Supplementary data is available at NAR Genomics & Bioinformatics online.

Conflict of interest

None declared.

Funding

Funding was provided in part by National Science Foundation (NSF) 2219900 (G.M.B.) awarded by the Division of Molecular and Cellular Biosciences and also by NSF grant 2243532 (G.M.B.) awarded by the Division of Integrative Organismal Systems.

Data availability

All datasets analyzed during the current study are listed and available through the NCBI SRA repository (https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP181693&o=acc_s%3Aa). FF.exe and TSV-Merger.exe files are available through the Borchert lab website (https://www.southalabama.edu/biology/borchertlab/downloads/FragmentFinder.zip) and all source codes are available through figshare (https://doi.org/10.6084/m9.figshare.29955488.v1) and GitHub (https://github.com/demeis117/FragmentFinder). Linux versions of FragmentFinder and TSV-Merger are also available through the Borchert lab website (https://www.southalabama.edu/biology/borchertlab/downloads/FragmentFinder_Linux.zip). Windows (or alternative Linux) executable availability: https://www.southalabama.edu/biology/borchertlab/.

References

  • 1. Ulitsky  I. Interactions between short and long noncoding RNAs. FEBS Lett. 2018;592:2874–83. 10.1002/1873-3468.13085. [DOI] [PubMed] [Google Scholar]
  • 2. Zhou  K, Diebel  KW, Holy  J  et al.  A tRNA fragment, tRF5-Glu, regulates BCAR3 expression and proliferation in ovarian cancer cells. Oncotarget. 2017;8:95377–91. 10.18632/oncotarget.20709. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Ender  C, Krek  A, Friedländer  MR  et al.  A human snoRNA with microRNA-like functions. Mol Cell. 2008;32:519–28. 10.1016/j.molcel.2008.10.017. [DOI] [PubMed] [Google Scholar]
  • 4. Alagia  A, Tereňová  J, Ketley  RF  et al.  Small vault RNA1-2 modulates expression of cell membrane proteins through nascent RNA silencing. Life Sci Alliance. 2023;6:e202302054. 10.26508/lsa.202302054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Olvedy  M, Scaravilli  M, Hoogstrate  Y  et al.  A comprehensive repertoire of tRNA-derived fragments in prostate cancer. Oncotarget. 2016;7:24766–77. 10.18632/oncotarget.8293. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Patterson  DG, Roberts  JT, King  VM  et al.  Human snoRNA-93 is processed into a microRNA-like RNA that promotes breast cancer cell invasion. NPJ Breast Cancer. 2017;3:25. 10.1038/s41523-017-0032-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Uchida  S, Bolli  R. Short and long noncoding RNAs regulate the epigenetic status of cells. Antioxid Redox Signaling. 2018;29:832–45. 10.1089/ars.2017.7262. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Yang  Y, Ishak Gabra  MB, Hanse  EA  et al.  MiR-135 suppresses glycolysis and promotes pancreatic cancer cell adaptation to metabolic stress by targeting phosphofructokinase-1. Nat Commun. 2019;10:809. 10.1038/s41467-019-08759-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Donovan  PD, McHale  NM, Venø  MT  et al.  tsRNAsearch: a pipeline for the identification of tRNA and ncRNA fragments from small RNA-sequencing data. Bioinformatics. 2021;37:4424–30. 10.1093/bioinformatics/btab515. [DOI] [PubMed] [Google Scholar]
  • 10. Griffiths-Jones  S, Grocock  RJ, van Dongen  S  et al.  miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acids Res. 2006;34:D140–4. 10.1093/nar/gkj112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Ohuchida  K, Mizumoto  K, Lin  C  et al.  MicroRNA-10a is overexpressed in human pancreatic cancer and involved in its invasiveness partially via suppression of the HOXA1 gene. Ann Surg Oncol. 2012;19:2394–402. 10.1245/s10434-012-2252-3. [DOI] [PubMed] [Google Scholar]
  • 12. Xu  Q, Li  P, Chen  X  et al.  miR-221/222 induces pancreatic cancer progression through the regulation of matrix metalloproteinases. Oncotarget. 2015;6:14153–64. 10.18632/oncotarget.3686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Wu  G, Li  Z, Jiang  P  et al.  MicroRNA-23a promotes pancreatic cancer metastasis by targeting epithelial splicing regulator protein 1. Oncotarget. 2017;8:82854–71. 10.18632/oncotarget.20692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Zhang  Y, Chen  D, Zhang  G  et al.  MicroRNA-23b-3p promotes pancreatic cancer cell tumorigenesis and metastasis via the JAK/PI3K and Akt/NF-κB signaling pathways. Oncol Lett. 2020;20:160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Liu  G, Shao  C, Li  A  et al.  Diagnostic value of plasma miR-181b, miR-196a, and miR-210 combination in pancreatic cancer. Gastroenterol Res Practice. 2020;2020:1. 10.1155/2020/6073150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Rao  X, Wan  L, Jie  Z  et al.  Upregulated miR-27a-3p indicates a poor prognosis in pancreatic carcinoma patients and promotes the angiogenesis and migration by epigenetic silencing of GATA6 and activating VEGFA/VEGFR2 signaling pathway. OTT. 2019;12:11241–54. 10.2147/OTT.S220621. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
  • 17. Duell  EJ, Lujan-Barroso  L, Sala  N  et al.  Plasma microRNAs as biomarkers of pancreatic cancer risk in a prospective cohort study. Intl J Cancer. 2017;141:905–15. 10.1002/ijc.30790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Chen  C, Demirkhanyan  L, Gondi  CS. The multifaceted role of miR-21 in pancreatic cancers. Cells. 2024;13:948. 10.3390/cells13110948. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Zhang  G, Liu  D, Long  G  et al.  Downregulation of microRNA-181d had suppressive effect on pancreatic cancer development through inverse regulation of KNAIN2. Tumour Biol. 2017;39:1010428317698364. [DOI] [PubMed] [Google Scholar]
  • 20. Pliatsika  V, Cherlin  T, Loher  P  et al.  MINRbase: a comprehensive database of nuclear- and mitochondrial-ribosomal-RNA-derived fragments (rRFs). Nucleic Acids Res. 2024;52:D229–38. 10.1093/nar/gkad833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Pliatsika  V, Loher  P, Magee  R  et al.  MINTbase v2.0: a comprehensive database for tRNA-derived fragments that includes nuclear and mitochondrial fragments from all The Cancer Genome Atlas projects. Nucleic Acids Res. 2018;46:D152–9. 10.1093/nar/gkx1075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Repetto  E, Lichtenstein  L, Hizir  Z  et al.  RNY-derived small RNAs as a signature of coronary artery disease. BMC Med. 2015;13:259. 10.1186/s12916-015-0489-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Alagia  A, Tereňová  J, Ketley  RF  et al.  Small vault RNA1-2 modulates expression of cell membrane proteins through nascent RNA silencing. Life Sci Alliance. 2023;6:e202302054. 10.26508/lsa.202302054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Martens-Uzunova  ES, Hoogstrate  Y, Kalsbeek  A. C/D-box snoRNA-derived RNA production is associated with malignant transformation and metastatic progression in prostate cancer. Oncotarget. 2015;6:17430–44. 10.18632/oncotarget.4172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Nakatsu  K, Jijiwa  M, Khadka  V  et al.  sRNAfrag: a pipeline and suite of tools to analyze fragmentation in small RNA sequencing data. Brief Bioinform. 2023;25:bbad515. 10.1093/bib/bbad515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Aparicio-Puerta  E, Lebrón  R, Rueda  A  et al.  sRNAbench and sRNAtoolbox 2019: intuitive fast small RNA profiling and differential expression. Nucleic Acids Res. 2019;47:W530–5. 10.1093/nar/gkz415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Rahman  RU, Gautam  A, Bethune  J  et al.  Oasis 2: improved online analysis of small RNA-seq data. BMC Bioinformatics. 2018;19:54. 10.1186/s12859-018-2047-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Liu  Q, Ding  C, Lang  X  et al.  Small noncoding RNA discovery and profiling with sRNAtools based on high-throughput sequencing. Brief Bioinform. 2021;22:463–73. 10.1093/bib/bbz151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Wan  C, Gao  J, Zhang  H  et al.  CPSS 2.0: a computational platform update for the analysis of small RNA sequencing data. Bioinformatics. 2017;33:3289–91. 10.1093/bioinformatics/btx066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Kuksa  PP, Amlie-Wolf  A, Katanic  Ž  et al.  SPAR: small RNA-seq portal for analysis of sequencing experiments. Nucleic Acids Res. 2018;46:W36–42. 10.1093/nar/gky330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Wu  X, Kim  TK, Baxter  D  et al.  sRNAnalyzer-a flexible and customizable small RNA sequencing data analysis pipeline. Nucleic Acids Res. 2017;45:12140–51. 10.1093/nar/gkx999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Shi  J, Ko  EA, Sanders  KM  et al.  SPORTS1.0: a tool for annotating and profiling non-coding RNAs optimized for rRNA- and tRNA-derived small RNAs. Genomics Proteomics Bioinformatics. 2018;16:144–51. 10.1016/j.gpb.2018.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Jeske  T, Huypens  P, Stirm  L  et al.  DEUS: an R package for accurate small RNA profiling based on differential expression of unique sequences. Bioinformatics. 2019;35:4834–6. 10.1093/bioinformatics/btz495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Hoogstrate  Y, Jenster  G, Martens-Uzunova  ES. FlaiMapper: computational annotation of small ncRNA-derived fragments using RNA-seq high-throughput data. Bioinformatics. 2015;31:665–73. 10.1093/bioinformatics/btu696. [DOI] [PubMed] [Google Scholar]
  • 35. Liao  Y, Smyth  GK, Shi  W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30:923–30. 10.1093/bioinformatics/btt656. [DOI] [PubMed] [Google Scholar]
  • 36. Tollervey  D, Kiss  T. Function and synthesis of small nucleolar RNAs. Curr Opin Cell Biol. 1997;9:337–42. 10.1016/S0955-0674(97)80005-1. [DOI] [PubMed] [Google Scholar]
  • 37. Hoagland  MB, Stephenson  ML, Scott  JF  et al.  A soluble ribonucleic acid intermediate in protein synthesis. J Biol Chem. 1958;231:241–57. 10.1016/S0021-9258(19)77302-5. [DOI] [PubMed] [Google Scholar]
  • 38. Johnston  R, Aldrich  A, Lyons  SM. Roles of ribosomal RNA in health and disease. Front RNA Res. 2024;1:1–23. 10.3389/frnar.2023.1331185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Koeppen  K, Hampton  TH, Barnaby  R  et al.  An rRNA fragment in extracellular vesicles secreted by human airway epithelial cells increases the fluoroquinolone sensitivity of P. aeruginosa. Am J Physiol: Lung Cell Mol Physiol. 2023;325:L54–65. 10.1152/ajplung.00150.2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Coley  AB, Stahly  AN, Kasukurthi  MV  et al.  MicroRNA-like snoRNA-derived RNAs (sdRNAs) promote castration-resistant prostate cancer. Cells. 2022;11:1302. 10.3390/cells11081302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Desgranges  E, Caldelari  I, Marzi  S  et al.  Navigation through the twists and turns of RNA sequencing technologies: application to bacterial regulatory RNAs. Biochim Biophys Acta, Gene Regul Mech. 2020;1863:194506. 10.1016/j.bbagrm.2020.194506. [DOI] [PubMed] [Google Scholar]
  • 42. Friedländer  MR, Chen  W, Adamidi  C  et al.  Discovering microRNAs from deep sequencing data using miRDeep. Nat Biotechnol. 2008;26:407–15. 10.1038/nbt1394. [DOI] [PubMed] [Google Scholar]
  • 43. Humphreys  DT, Suter  CM. miRspring: a compact standalone research tool for analyzing miRNA-seq data. Nucleic Acids Res. 2013;41:e147. 10.1093/nar/gkt485. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Hackenberg  M, Rodríguez-Ezpeleta  N, Aransay  AM. miRanalyzer: an update on the detection and analysis of microRNAs in high-throughput sequencing experiments. Nucleic Acids Res. 2011;39:W132–8. 10.1093/nar/gkr247. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

lqag017_Supplemental_Files
lqag017_supplemental_files.zip (1.6MB, zip)

Data Availability Statement

All datasets analyzed during the current study are listed and available through the NCBI SRA repository (https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP181693&o=acc_s%3Aa). FF.exe and TSV-Merger.exe files are available through the Borchert lab website (https://www.southalabama.edu/biology/borchertlab/downloads/FragmentFinder.zip) and all source codes are available through figshare (https://doi.org/10.6084/m9.figshare.29955488.v1) and GitHub (https://github.com/demeis117/FragmentFinder). Linux versions of FragmentFinder and TSV-Merger are also available through the Borchert lab website (https://www.southalabama.edu/biology/borchertlab/downloads/FragmentFinder_Linux.zip). Windows (or alternative Linux) executable availability: https://www.southalabama.edu/biology/borchertlab/.

Articles from NAR Genomics and Bioinformatics are provided here courtesy of Oxford University Press

RESOURCES