Detecting alternatively spliced transcript isoforms from single‐molecule long‐read sequences without a reference genome

Abstract

Alternative splicing (AS) is a major source of transcript and proteome diversity, but examining AS in species without well-annotated reference genomes remains difficult. Research on both human and mouse has demonstrated the advantages of using Iso-Seq™ data for isoform-level transcriptome analysis, including the study of AS and gene fusion. We applied Iso-Seq™ to investigate AS in Amborella trichopoda, a phylogenetically pivotal species that is sister to all other living angiosperms. Our data show that, compared with RNA-Seq data, the Iso-Seq™ platform provides better recovery on large transcripts, new gene locus identification and gene model correction. Reference-based AS detection with Iso-Seq™ data identifies AS within a higher fraction of multi-exonic genes than observed for published RNA-Seq analysis (45.8% vs. 37.5%). These data demonstrate that the Iso-Seq™ approach is useful for detecting AS events. Using the Iso-Seq-defined transcript collection in Amborella as a reference, we further describe a pipeline for detection of AS isoforms from PacBio Iso-Seq™ without using a reference sequence (de novo). Results using this pipeline show a 66%-76% overall success rate in identifying AS events. This de novoAS detection pipeline provides a method to accurately characterize and identify bona fide alternatively spliced transcripts in any nonmodel system that lacks a reference genome sequence. Hence, our pipeline has huge potential applications and benefits to the broader biology community.

De novo assembly and characterization of the Hucho taimen transcriptome

“…We then searched for duplicate genes using the reciprocal self-BLAST hit method. After filtering out gene variations (alternative splicing or allele) using a de novo alternative splice detection pipeline (Liu et al 2017), a reciprocal self-BLAST search for all protein sequences was performed to identify putative paralog pairs, which were further filtered out with identities >70% and alignment coverage >50%. The protein sequences of paralog pairs were aligned with MUSCLE, and the protein alignments were back-translated to coding sequence (CDS) alignments using TreeBest software.…”

Pathogenic impact of transcript isoform switching in 1209 cancer samples covering 27 cancer types using an isoform-specific interaction network

“…Nevertheless, as Kallisto only quantifies known transcripts, we might have underestimated the impact of altered alternative splicing by not considering novel transcripts. Long-read sequencing (Tilgner et al, 2015) in bulk or on single cells (Gupta et al, 2018;X. Liu et al, 2017) are ideal methods to overcome these problems.…”

Clustering de Novo by Gene of Long Reads from Transcriptomics Data

“…Non model species require de novo approaches, thus two bioinformatics tools dedicated to them have emerged so far [42,41]. Both comprise a pipeline conceived to process Pacific Biosciences Isoseq [3] reads only and require high quality long reads.…”

“…Clustering problems applied to the specificity of long reads start to emerge. Such needs were already of concern in the past long read literature [11,41] and are even more acute when a mapping strategy cannot be taken into consideration. We place ourselves in the particular framework of de novo identification.…”

“…While several works based on long read mapping on a reference were proposed, methodological contributions that would enable to make the most of this promising data remain rare in particular for non model species. To our knowledge, two contributions [42,41] propose respectively to de novo detect alternative variants and to cluster and detect isoforms in long reads transcriptome datasets. However these tools highly rely on the property of high accuracy proposed by Pacific Biosciences Consensus Circular Sequence (CCS) long reads, thus do not apply to Oxford Nanopore Technologies reads.…”