Full-length transcriptome sequencing reveals extreme incomplete annotation of the goat genome.

Abstract

Despite recent advances in generating high-quality reference genome assemblies, the genome sequences for most livestock species, including goats, are still poorly annotated. Single-molecule long-read sequencing has greatly facilitated gene annotation by obtaining full-length transcripts. In this study, we generated full-length transcriptome data for samples from abomasum (n=2) and testicle (n=1), using PacBio Iso-Seq technology. We further combined these data with published data from abomasum (5ZY, SRR8618141) to evaluate and improve the gene annotation of the goat genome. We identified 14.5-16.3% of novel genes per sample from the four Iso-Seq datasets. At the transcript level, 40.6% of them were novel, including 29.7% novel transcripts from known genes and 10.9% from novel genes. We further verified the expression of novel genes in four additional RNA-seq data and found that the expression level of novel genes was significantly lower than that of known genes, indicating that the lowly expressed genes tend to be missed in the current genome annotation. This study shows the superiority of full-length transcriptome data in gene annotation, and more such data are required to improve the gene annotation for goat genome and other species.