Purchase this article with an account.
Terry Gaasterland, Lee E. Edsall, Robert N. Weinreb, Kaweh Mansouri, Kang Zhang, Douglas E. Gaasterland, Michael A. Hauser, Julia E. Richards, Janey L. Wiggs, NEIGHBOR Consortium Investigators; Parameterized Analysis of Weighted Variants in Blended Cohorts Optimizes Analysis of Exomes from Primary Open Angle Glaucoma Cases from the NEIGHBOR Study. Invest. Ophthalmol. Vis. Sci. 2012;53(14):4516.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Develop a logic-based system to analyze damaging variants and maximize discovery of disease-related genome variations identified through DNA sequencing of captured exons genome-wide (exome).
400 primary open angle glaucoma (POAG) cases were selected for sequencing from 2,517 POAG cases in the NEIGHBOR GWAS study, all with characteristic visual field defects consistent with glaucomatous optic neuropathy on reliable tests; additional selection criteria for sequencing gave priority to family history of disease and elevated IOP (>22mmHg). In addition, the affected close relatives of 18 NEIGHBOR cases and 12 individuals of African background were included. The blended cohort of POAG cases included five classes: unrelated individuals with family history (FH); unrelated individuals with unknown FH; pairs of siblings or cousins; non-Caucasian background.Commercial exome capture kits were applied to hybridize probes to target DNA in solution. For each person, captured DNA was sequenced (Illumina HiSeq) generating >50 million pairs of 100-base reads for 340,333 exons with 36,172,938 bases. Reads were mapped to reference genome hg19 and SNPs (single nucleotide polymorphisms) identified using published methods.SNPs in or near genes were annotated with regard to location, effect on encoded protein, and effect on regulatory elements.
Sequence data from 217 exomes yielded over 5 million coding SNPs (cSNP) different from hg19, averaging 23,274(±813) cSNPs per exome. 92% of targeted bases had >9x coverage, and 96%>1x. cSNPs included 11,686(±376) missense, 229(±8) nonsense, and 59(±4) splice-site changes average per exome. Each SNP in each individual was assigned a weight based on zygosity, gene disruption, general population frequency, frequency in ethnic backgrounds (weighted by reported disease frequency), FH of individual, affected relative sharing, and participation in pathways with variants from other cases (a reasoning feedback loop). A decision tree system was configured to accept a set of weighting parameters and compute weight for each SNP instance. Genes were then ranked by aggregated weighted cohort variants. The decision tree built for the first 217 exomes will be extended to the entire dataset of 400 exomes.
Our parameterized weighted variant analysis fills an unmet need in exome sequencing. It ranks disruptive SNPs across a cohort and ranks genes based on SNP weights. Implemented efficiently with a deductive inference algorithm, the method allows investigators to explore hypotheses about the effect on disease of weighting parameters. It enhances potential to discover disease-related genes.
This PDF is available to Subscribers Only