《SOAPindel_Paper_20160306》.pdfVIP

SOAPindel: Efficient identification of indels from short paired reads Shengting Li1,2, Ruiqiang Li1#, Heng Li6, Jianliang Lu1, Yingrui Li1, Lars Bolund1,3, Mikkel H. Schierup2*, Jun Wang1,4,5* 1. BGI Shenzhen, Shenzhen 518000, China 2. Bioinformatics Research Centre, Aarhus University, DK 8000 Aarhus C., Denmark 3. Human Genetics, Aarhus University, DK 8000 Aarhus C, Denmark 4. The Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen. 5. Department of Biology, University of Copenhagen, Copenhagen, Denmark. 6. Broad Institute, 7 Cambridge Center, MA 02142, USA # Present address: Biodynamic Optical Imaging Center, and College of Life Sciences, Peking University, Beijing 100871, China * Corresponding authors: wangj@, mheide@birc.au.dk Abstract We present a new approach to calling indels from paired end, short read sequencing data. The method performs extensive local assembly of reads in regions of the genome where a large fraction of partners of mapped read does not map. Such inconsistencies are presumably due to a disruption of the read by an indel. Indels are subsequently called in these regions when a local assembly path at some point anchors to the reference sequence, allowing heterozygotes to be efficiently called. The approach is implemented in the program SOAPindel. We 1 demonstrate, using simulations, data analysis and validation, that the method possesses a high sensitivity and specificity even in case of relatively long indels (20-200 bp), where other methods generally lack sensitivity. Introduction Indel calling from mapping short paired-end sequences to a reference genome remains a difficult problem because mapping per se rarely allows calling of indels with size longer than 25% of the read length (Li, Ruan et al. 2008; Li and Du