Skip to main content
. 2020 Jan 14;9:e49658. doi: 10.7554/eLife.49658

Figure 3. Cap-dependent ligation of an adapter enables detection of authentic RNA 5′ ends.

(A) 5′ adapter RNA ligation reduces 3′ bias in nanopore DRS data at RCA (AT2G39730) from 0.92 to 0.03. Blue line, exonic read coverage at RCA for reads without (left) and with adapter (right); orange line, median coverage; orange shaded area, interquartile range (IQR). Change in 3′ bias can be measured using the IQR/median = quartile coefficient of variation (QCoV). (B) 5′ adapter RNA ligation reduces 3′ bias in nanopore DRS data. Histogram showing the QCoV in per base coverage for each gene in the Araport11 annotation, for reads with a 5′ adapter RNA (orange), and reads without a 5′ adapter RNA (blue). (C) Cap-dependent adapter ligation allows identification of authentic 5′ ends using nanopore DRS. The cumulative distribution function shows the distance to the nearest Transcription Start Site (TSS) identified from full-length transcripts cloned as part of the RIKEN RAFL project (left), or 5’ tag identified from nanoPARE data (right), for reads with a 5′ adapter RNA (orange) compared with reads without a 5′ adapter RNA (blue). (D) Cap-dependent adapter ligation enabled resolution of an additional 11 nt of sequence at the RNA 5′ end. Histogram showing the nucleotide shift in the largest peak of 5′ coverage for each gene in data obtained using protocols with vs without a 5′ adapter RNA ligation. (E) For RCA (AT2G39730), the 5′ end identified using cap-dependent 5′ adapter RNA ligation protocol was consistent with Illumina RNAseq and full-length cDNA start site data but differed from the 5′ ends in the Araport11 and AtRTD2 annotations. Grey, Illumina RNAseq coverage; blue, nanopore DRS 5′ end coverage generated without a cap-dependent ligation protocol; green/pink, nanopore DRS 5′ end coverage for read alignments generated using the cap-dependent ligation protocol with (green) and without (pink) 5′ adapter RNA; orange, 5’ coverage of nanoPARE data; blue, TSSs identified from full-length transcripts cloned as part of the RIKEN RAFL project; black, Araport11 and AtRTD2 annotations (with duplicated 5’ positions removed).

Figure 3—source data 1. Transcriptional start site tags from RIKEN cDNA clones – Figure 3C.
Figure 3—source data 2. Transcriptional start site tags from nanoPARE sequencing – Figure 3C.

Figure 3.

Figure 3—figure supplement 1. Nanopore DRS with cap-dependent ligation of 5′ adapter RNA.

Figure 3—figure supplement 1.

(A) Histogram showing the distribution of 5′ adapter RNA length in the nanopore raw current signal, as inferred from alignment of the mRNA sequence to the signal using nanopolish eventalign. The median signal length was 1441 points and 96% of adapter signals were 3000 points or less. (B) Out-of-bag receiver operator characteristic curve showing the performance of the trained convolutional neural network at detecting 5′ adapter RNA using 3000 points of signal. The curve was generated using five-fold cross validation. (C) Out-of-bag precision recall curve showing the performance of trained neural network, generated using five-fold cross validation. (D) Alternative transcription start sites were identified using nanopore DRS with cap-dependent ligation of a 5′ end adapter at the AT1G17050 and AT5G18650 genes. Orange, 5’ coverage for capped nanoPARE reads; blue track, nanopore DRS coverage with cap-dependent ligation of 5′ adapter RNA; blue, isoforms detected by nanopore DRS with cap-dependent ligation of 5′ adapter RNA; black, Araport11 annotation. (E) Reads mapping to ERCC RNA spike-ins lack approximately 11 nt of sequence at the 5′ end. Histogram showing the distance to the 5′ end for ERCC RNA spike-in reads (each spike-in is shown in a different colour; only those with > 1000 supporting reads are shown). (F) Reads mapping to in vitro transcribed mGFP lack approximately 11 nt of sequence at the 5′ end. Histogram showing the distance to the 5′ end for in vitro transcribed mGFP. (G) Araport11 annotation overestimates the length of 5′ UTRs. The cumulative distribution function shows the distance to the nearest TSS identified from full-length transcripts cloned as part of the RIKEN RAFL project (blue) and Araport11 annotation (orange). (H) Nanopore DRS detects miR170/miR171 cleavage products of Hairy Meristem 1 (HAM1, AT2G45160) transcripts. Orange, 5’ coverage from capped nanoPARE reads; purple, 5’ coverage from uncapped nanoPARE reads; blue, nanopore DRS 5’ coverage; grey, miRNA target site alignment is shown; black, Araport11 annotation.
Figure 3—figure supplement 1—source data 1. microRNA cleavage site predictions supported by enrichment of nanopore 5’ ends – Figure 3—figure supplement 1H.