PgmNr 107: Identification of low-level parental somatic mosaic SNVs and InDels in a large exome sequencing cohort of individuals with Mendelian disorders.Authors:
T. Gambin 1; J.A. Karolak 1; Q. Liu 1; S.N. Jhangiani 2; Z.H. Coban Akdemir 1; J.R. Lupski 1,2,3,4; P. Stankiewicz 1
View Session Add to Schedule
1) Dept. of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX; 2) Human Genome Sequencing Center, Baylor College of Medicine, Houston, TX; 3) Texas Children's Hospital, Houston, TX; 4) Dept. of Pediatrics, Baylor College of Medicine, Houston, TX
Mosaic variants due to somatic mutations can cause both cancer and Mendelian conditions. In families with children affected with a Mendelian disorder, detection of somatic mosaicism in the parental samples is essential to better estimate the recurrence risk. Although high depth next generation sequencing facilitates detection of low-level mosaic variants, thus far, only a few studies have reported results of systematic analyses of low-level mosaicism in large exome sequencing (ES) cohorts. We computationally analyzed ES data from 882 unrelated family trios (with the complete VCF & BAM files) recruited in the Baylor Hopkins Center for Mendelian Genomics (BHCMG) at BCM. After removal of the low quality samples and variants with minor allele frequency (MAF) >0.0001 in the 1000Genomes or local databases, we found 3,156 apparent de novo SNVs or indels in 768 probands. To identify the candidate low-level (<10%) somatic mosaic variants, we calculated alternate allele fraction (AAF) for all rare (<0.0001) variants in the entire trio dataset with the total depth of read coverage >20x and AAF between 0.3-0.7 in the affected patients (likely heterozygous variant), and AAF lower than 0.1 but greater than zero (likely low-level mosaic variant) in one of the parental samples. We found 76 variants in 67 unrelated family trios fulfilling these criteria, including 53 missense, 9 splicing, 5 indels, 4 stop-gain, 4 non-coding, and 1 synonymous SNVs. To assess the quality of these findings, for each genomic position of putative mosaic SNVs, we retrieved the pileup information from 7788 ES samples. We found that on average 4% of BHCMG samples (ranging from 0 to 45%) carry at least one read supporting alternative allele at the position of potential mosaic variant. The selected candidate mosaic variants are being validated using three highly sensitive orthogonal molecular assays: (i) PCR amplicon-based next generation sequencing with high coverage; (ii) droplet digital PCR (ddPCR); and (iii) blocker displacement amplification (BDA) (PMID: 29805844). Moreover, we are attempting to evaluate and identify the most reliable predictors (e.g. GC content, mapping quality, frequency of individuals with alternate reads) that can facilitate discrimination between true positive from false positive mosaic variants. These predictors will be used for building a classification tool enabling accurate detection of low-level mosaicism in ES samples.