High resolution annotation of Zebrafish transcriptome using long-read sequencing

With the emergence of zebrafish as an important model organism, a concerted effort has been made to study its transcriptome. This effort is limited by gaps in zebrafish annotation, which is especially pronounced concerning transcripts dynamically expressed during zygotic genome activation (ZGA). To date, short read sequencing has been the principal technology for zebrafish transcriptome annotation. In part because these sequence reads are too short for assembly methods to resolve the full complexity of the transcriptome, the current annotation is rudimentary. By providing direct observation of full-length transcripts, recently refined long-read sequencing platforms can dramatically improve annotation coverage and accuracy. Here, we leveraged the SMRT platform to study the early ZGA-stage zebrafish transcriptome. Our analysis revealed additional novelty and complexity in the zebrafish transcriptome, identifying 2748 high confidence novel transcripts that originated from previously unannotated loci and 1835 new isoforms in previously annotated genes. Pooled RNA of α-amanitin / untreated embryos were collected and profiled with long-read sequencing. Temporally corresponding pre/post ZGA pooled embryonic RNA samples were profiled with short-read RNA-seq. Long-read raw data were assembled into transcripts using IsoSeq  (PMID: 27407110), mapped to the reference GRCz10 genome using GMAP [PMID:15728110] and annotated against the reference transcriptome using Cuffcompare [PMC3334321]. Novel transcripts were compared to short-read data and computationally validated in constructing a final long-read augmented transcriptome.