Improving Porcine Transcriptome Annotation By Integrating EST, mRNAs and Whole Blood RNA-Seq Data

The pig genome annotation (Ensembl Sscrofa 10.2) has been widely used as a reference in genome-scale studies. However, except for ~2,000 manually annotated genes, most genes have only automated annotation, with <1.2 isoforms per coding gene on average. In recent RNA-seq data analysis, we found >20% of uniquely mapped reads couldn’t be assigned to gene features due to ambiguity or no corresponding features. Visualizing BAM files revealed many genes with incorrectly annotated untranslated regions (UTR). To better mine RNA-seq data, the objective here was to improve pig genome annotation. First, 1,592,114 cleaned ESTs and mRNA sequences from GenBank were used to improve pig genome annotation by using Program to Assemble Spliced Alignments (PASA). On average, 1.7 isoforms were added to each of 10,011 protein-coding genes, and 28,218 potential polyA sites were deduced. Next, genome-guided transcriptome assembly and de novo transcriptome assembly were performed with 65 Million (M) paired-end and 62M single-end Illumina RNA-seq reads from a single pig blood sample. A comprehensive blood transcriptome was built by using PASA. Visualizing the assembled transcriptome along with BAM files indicated there were many novel genes, as well as isoforms, corrected UTR, novel transcription start sites, and evidence for incorrect annotation for existing Ensembl genes. We are further annotating the assembled transcriptome by BLASTN, BLASTX, and TransDecoder and manually by WebApollo. In the future, we will integrate these data with public high quality RNA-seq data to assemble a more comprehensive pig transcriptome to improve the pig genome annotation. Funded by USDA-NIFA Grant #2011-68004-30336.