Allelic-Imbalance Analysis in Pooled RNA-Seq Samples Identifies Muscle-Associated Genetic Markers in Rainbow Trout: Improved Bioinformatics Practices

Coding SNPs (cSNPs) are likely to change the biological function of a protein. We have demonstrated that assessment of phenotype-associated allelic imbalances in pooled RNA-Seq samples is a fast means of identifying “large-effect” genetic markers. In this study, we compared two main variant calling bioinformatics pipelines, GATK and SAMtools, for discovery of cSNPs with allelic-imbalances that are associated with total-body weight, muscle yield, fat content, shear force and whiteness. Phenotypic data were collected from 500 fish representing 98 families (5 fish/family) of a growth-selected line.  RNA-Seq was used to sequence the muscle transcriptome from 26 families showing divergent phenotypes (4 high- versus 4 low-ranked families/each trait).  GATK detected 468,494 putative cSNPs: 6,228 showed allelic imbalances (>1.4 as an amplification and <0.7 as loss of heterozygosity). SAMtools detected 188,995 putative cSNPs; 943 had significant allelic imbalances between the low and high families.  A total of 648 cSNPs with allelic imbalances were common between the two datasets, indicating significant differences in algorithms. A total of 6,966 non-redundant cSNPs were identified in the two datasets; 3,904 mapped to 1,606 protein-coding genes (with 13.5% non-synonymous cSNPs), and 1,279 mapped to 588 lncRNAs. Twenty cSNPs mapped to 6 genes showing differential phenotypic expression. Optimization of GATK and SAMtools to fit RNA-Seq pooled-sample analysis will be presented. These results will be used to build a cSNP-chip for genomic selection of muscle growth and quality traits in rainbow trout.