Muver, a computational framework for accurately calling accumulated mutations

Adam B Burkholder,Christopher A Lavender,Scott A Lujan,Sara A Grimm,David C Fargo,Thomas A Kunkel

doi:10.1186/s12864-018-4753-3

Adam B Burkholder, Christopher A Lavender + Show 4 more

Open Access

https://doi.org/10.1186/s12864-018-4753-3

Copy DOI

Abstract

BackgroundIdentification of mutations from next-generation sequencing data typically requires a balance between sensitivity and accuracy. This is particularly true of DNA insertions and deletions (indels), that can impart significant phenotypic consequences on cells but are harder to call than substitution mutations from whole genome mutation accumulation experiments. To overcome these difficulties, we present muver, a computational framework that integrates established bioinformatics tools with novel analytical methods to generate mutation calls with the extremely low false positive rates and high sensitivity required for accurate mutation rate determination and comparison.ResultsMuver uses statistical comparison of ancestral and descendant allelic frequencies to identify variant loci and assigns genotypes with models that include per-sample assessments of sequencing errors by mutation type and repeat context. Muver identifies maximally parsimonious mutation pathways that connect these genotypes, differentiating potential allelic conversion events and delineating ambiguities in mutation location, type, and size. Benchmarking with a human gold standard father-son pair demonstrates muver’s sensitivity and low false positive rates. In DNA mismatch repair (MMR) deficient Saccharomyces cerevisiae, muver detects multi-base deletions in homopolymers longer than the replicative polymerase footprint at rates greater than predicted for sequential single-base deletions, implying a novel multi-repeat-unit slippage mechanism.ConclusionsBenchmarking results demonstrate the high accuracy and sensitivity achieved with muver, particularly for indels, relative to available tools. Applied to an MMR-deficient Saccharomyces cerevisiae system, muver mutation calls facilitate mechanistic insights into DNA replication fidelity.

Highlights

Identification of mutations from next-generation sequencing data typically requires a balance between sensitivity and accuracy
We introduce muver, a mutation calling framework that streamlines the use of common tools and applies novel methods to leverage the power of matched progeny and ancestor to minimize the false positive rate (FPR) without sacrificing sensitivity
Mutation accumulation experiments Descendant samples are compared to ancestral samples to detect mutations that accumulate over the course of the experiment and to exclude changes that occur during strain construction and handling

Summary

Introduction

Identification of mutations from next-generation sequencing data typically requires a balance between sensitivity and accuracy This is true of DNA insertions and deletions (indels), that can impart significant phenotypic consequences on cells but are harder to call than substitution mutations from whole genome mutation accumulation experiments. To overcome these difficulties, we present muver, a computational framework that integrates established bioinformatics tools with novel analytical methods to generate mutation calls with the extremely low false positive rates and high sensitivity required for accurate mutation rate determination and comparison. Decades of in vivo mechanistic insights into these processes came from mutation rates derived from reporter gene assays [14]. Long repeat tracts are rare in exons of protein coding genes [18], leading to underestimation of whole genome indel rates

Methods

Results

Discussion

Conclusion