Beyond SNVs: comprehensive cohort-level analytics of 100 samples from the eyeGENE® program

Abstract

Purpose : Next-generation sequencing provides copious gene- and nucleotide-level data, allowing for additional analysis of disease-causing variation beyond Sanger sequencing. Inherited retinal degeneration (IRD) has been associated with variation in over 300 genes to date, and a major limitation is sorting pathogenic variants from those of uncertain significance. Here, we utilize deep sequencing to provide intra- and inter-sample metrics for automated filtering and prioritization of single nucleotide variants (SNV), insertions-deletions (indels), structural variants (SV), copy number variation (CNV), and sex determination.

Methods : Custom capture for 147 target genes for IRD was performed using Nimblegen SeqCap EZ choice (Roche) to generate libraries from 100 samples from the eyeGENE® program with clinical diagnosis of IRD, which were sequenced using MiSeq (Illumina). Alignment with BWA and variant calling for SNVs and indels with GATK was followed by annotation, variant filtering, and prioritization using Variant Effect Predictor and Gemini. CNV and SV callings utilized CoNVaDING, CREST and others, and sex determination was performed using autosomal normalization of X chromosome variation excluding the opsin locus.

Results : Target gene coverage was nearly 98% with average depth of >500x. Automated variant filtering and prioritization correlated genotype to pre-test diagnosis in nearly half of patients. Large scale CNVs were detected in 11% and small-scale deletions of 53-259bp in 6% of cases, and providing direct evidence for molecular diagnosis in an additional 8%. This includes a previously reported 353bp Alu insertion in the MAK gene and a 36.4 kb deletion in the TRPM1 gene. Accurate sex determination was possible in 99% of samples.

Conclusions : Comprehensive analytics of next-generation sequencing data provide information beyond single- or oligo-nucleotide variation by normalization across both target genes and samples. Pathogenic CNVs and structural variants are present in a significant number of samples. Cross-checking patient demographics, phenotypic information, and DNA library and sequencing quality with NGS data are important quality controls for clinical diagnostics.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.