PEMer: a computational framework with simulation-based error models for inferring genomic structural variants from massive paired-end sequencing data

Jan O Korbel,Michael Snyder,Xinmeng Mu,Philip Cayting,Nicholas Carriero,Mark B Gerstein,Zhengdong Zhang,Alexej Abyzov

doi:10.1186/gb-2009-10-2-r23

Abstract

Paired-End Mapper (PEMer) enables mapping of genomic structural variants at considerably enhanced sensitivity, specificity and resolution over previous approaches.

Highlights

Following the sequencing of the genomes of hundreds of species over the past years, measuring variation within individuals of a species - such as across human beings - has become a center of attention in genomics [1]
Paired-End Mapper (PEMer) implements a number of subsequent computational procedures, or steps, which have been developed as a set of modular components
When processing relatively short sequenced ends we recommend novel read-indexing approaches that directly compensate for variation in the mappability of short sequences in the context of a complex, repeat-rich reference genome [28,29]

Summary

Introduction

Following the sequencing of the genomes of hundreds of species over the past years, measuring variation within individuals of a species - such as across human beings - has become a center of attention in genomics [1]. [10,11]), scoring high-density microarray platforms [12,13,14,15], measuring DNA sequence read densities [16], detecting split sequence reads [17], or comparing different human genome assemblies [18,19,20] Each of these complementary approaches enables the identification of at least a subset of SVs at a reasonable confidence level. Experimental validations indicated a reasonably successful performance of these approaches, a suitably parameterized approach to SV calling will be necessary to generate high confidence SV sets and to optimize the specificity and sensitivity of SV calling In this regard it is evident that future studies that will utilize dense maps of structural variation in the genome for associating SV genotypes with phenotypic data will rely on high-confidence methods for SV calling. PEMer can be downloaded from [27], where instructions on how to install the framework are provided

Methods

Results

Conclusion