Rare Variant Analysis of Next Generation Sequencing Studies with the Variable Binning / Variable Threshold (VB/VT) Collapsing Algorithm
Final Report Abstract
Region-based genome-wide scans are usually performed by use of a priori chosen analysis regions. Such approaches will likely miss the region comprising the strongest signal and, thus, may result in increased type II error rates and decreased power. We developed the genomic exhaustive collapsing scan (GECS) to allow for an exhaustive scan of all possible contiguous genomic regions, or bins, with the collapsing test and eliminates the prior choice of candidate bins. Instead, the space of all possible bins is explored and tested. This eliminates binning as a source of increased type II errors and improves power to detect trait-related genetic associations under a variety of disease etiology scenarios. Furthermore, the speed-up by several orders of magnitude allows for computation of non-conservative genome-wide significance thresholds by use of permutation, thereby allowing control of the family-wise error rate for genome-wide analyses, leading to improved power when compared to more conservative correction methods such as Bonferroni’s. We show that GECS is powerful in detecting phenotypic associations of genomic regions harboring rare variants and for refining our understanding of their contribution to predisposing to complex diseases. Our implementation allows for the identification of regions comprising the strongest signals in large, whole-genome rare-variant association studies and is publicly available at https://github.com/ddrichel/GECS. We conclude that GECS is well suited for whole-genome and whole-exome association analyses. Application of exhaustive scans is not limited to association testing and could be useful in further applications, in particular for studying methylation and evolutionary selection.
Publications
-
(2019) The exhaustive genomic scan approach, with an application to rare-variant association analysis
George Kanoungi, Michael Nothnagel, Tim Becker, Dmitriy Drichel
