Structural Variation Detection with Read Pair Information: An Improved Null Hypothesis Reduces Bias

• DOI: 10.1089/cmb.2016.0124
• Publication Date: 2016-09-28T20:01:50Z
• Authors: Kristoffer Sahlin; Mattias Frånberg; Lars Arvestad
• Keywords: Fragment (logic); Null (SQL); Null model; Variation (astronomy); Statistical power; Structural Variation

Reads from paired-end and mate-pair libraries are often utilized to find structural variation in genomes, one common approach is use their fragment length for detection. After aligning read pairs the reference, pair distances analyzed statistically significant deviations. However, previously proposed methods based on a simplified model of observed lengths that does not agree with data. We show how this limits statistical analysis identifying variants propose new by adapting we have introduced contig scaffolding, which agrees From model, derive an improved null hypothesis when applied variant caller CLEVER, reduces number false positives corrects bias contributes more deletion calls than insertion calls. advise developers callers length-based adapt concepts our hypothesis.