HyDRA: a pipeline for integrating long- and short-read RNAseq data for custom transcriptome assembly

Abstract

BackgroundShort-read RNA sequencing (RNAseq) has widely been used to sequence RNA from a wide range of different tissues, developmental stages and species. However, the technology is limited by inherent biases and its inability to capture full-length transcripts. Long-read RNAseq overcomes these issues by providing reads that can span multiple exons, resolve complex repetitive regions and the capability to cover entire transcripts. Unfortunately, this technology is still prone to higher error rates. Noncoding RNA transcripts are highly specific to different cell types and tissues and remain underrepresented in current reference annotations. This problem is exacerbated by the dismissal of sequenced reads that align to genomic regions that do not contain annotated transcripts, resulting in approximately half of the expressed transcripts being overlooked in transcriptional studies. ResultsWe have developed a pipeline, named HyDRA (Hybrid de novo RNA assembly), which combines the precision of short reads with the structural resolution of long reads, enhancing the accuracy and reliability of custom transcriptome assemblies. Deep, short- and long-read RNAseq data derived from ovarian and fallopian tube samples were used to develop, validate and assess the efficacy of HyDRA. We identified more than 50,000 high-confidence long noncoding RNAs, most of which have not been previously detected using traditional methods. ConclusionsHyDRAs assembly performed more than 40% better than a similar assembly obtained with the top-ranked stand-alone de novo transcriptome short-read-only assembly tool and over 30% better than one obtained with the best-in-class multistep short-read-only approach. Although long-read sequencing is rapidly advancing, the vast availability of short-read RNAseq data will ensure that hybrid approaches like the one implemented in HyDRA continue to be relevant, allowing the discovery of high-confidence transcripts within specific cell types and tissues. As the practice of performing hybrid de novo transcriptome assemblies becomes commonplace, HyDRA will advance the annotation of coding and noncoding transcripts and expand our knowledge of the noncoding genome.