Comparison and Characterisation of Mutation Calling from Whole Exome and RNA Sequencing Data

Whole exome sequencing has had low uptake in livestock species, despite allowing accurate analysis of single nucleotide variant (SNV) mutations. Transcriptomic data in the form of RNA sequencing has been generated for many livestock species and also represents a source of mutational information. However, there is little information on the accuracy of using this data for the identification of SNVs.

Researchers at the University College Dublin generated a bovine exome capture design and used it to sequence and call mutations from a lactating dairy cow model genetically divergent for fertility (Fert+, n=8; Fert-, n=8). They compared mutations called from liver and muscle transcriptomes from the same animals. Their exome capture demonstrated 99.1% coverage of the exome design of 56.7MB, whereas transcriptomes covered 55 and 46.5% of the exome, or 24.4 and 20.7MB, in liver and muscle respectively after filtering. The researchers found that specificity of SNVs in the transcriptome data is approximately 75% following basic hard-filtering, and could be increased to above 80% by increasing the minimum threshold of reads covering SNVs, but this effect was negated in more highly covered SNVs. RNA-DNA differences, SNVs found in transcriptome but not exome, were discovered and shown to have significantly increased levels of transition mutations in both tissues. Functional annotation of non-synonymous SNVs specific to the high and low fertility phenotypes identified immune response-related genes, supporting previous work that has identified differential expression in the same genes. Publically available RNAseq data may be analysed in a similar way to further increase the utility of this resource.

MDS plot was constructed using the combined VCF