Searching for Sequencing Signal Anomalies Associated with Genomic Structural Variations

Abstract

Genomic structural variations (SVs) are among the main sources of genetic diversity. Structural variants as mutagens may significantly affect human health, causing hereditary diseases and cancers. Existing methods analyze high-throughput sequencing data to find structural variants. Despite substantial progress in their development, the methods still fail to detect structural variations with an accuracy sufficient for their use in diagnosis. Analysis of the sequencing coverage signal (i.e., the number of aligned sequencing reads for every point of a genome) holds the new potential for designing approaches to structural variation detection and can be used as time-series analysis. A method to detect repetitive patterns in the coverage signal was developed based on the time series-assessing algorithms KNN (K-nearest neighbor) and SAX (Symbolic Aggregation Approximation). Using the rich dataset encompassing the full genomes of 911 individuals with different ethnic backgrounds from the Human Genome Diversity Project, generalized patterns of the coverage signal were constructed for regions in the vicinity of breakpoints corresponding to various structural variant types. The patterns were used to develop a software package for fast detection of anomalies in the coverage signal.