Bootstrap-Based Quality Scores for NGS Variant Calling

Abstract

Variant calling, the identification of DNA sites that differ from a reference sequence using Next Generation Sequencing (NGS) methods is still an inherently error prone process. Quality scores of variant calling results in NGS studies is a key utility for identifying and prioritizing variants for validation in clinical practice. We investigated the robustness of a wide range of quality scores using bootstrap resampling of the short reads resulting from an NGS exome sequencing study, and evaluated their use as additional quality scores to characterize the called variants. Besides quality scores from the measurement process itself and the biological and clinical relevance of a variant, the bootstrap derived quality scores provide an additional information source for validation prioritization. Additionally, our results indicate possibilities for decreasing the depth of the NGS measurement and increasing the robustness of bioinformatic post-processing.