Abstract

High throughput sequencing provides access to important information on genes, their function and genetic variation on genomes. One way of post processing is a reference guided assembly. For this approach, one needs an evolutionary closely related reference genome. However, in the end of 2009, only a few completely referenced organisms are available. Therefore, we have also considered more distantly related organisms as reference candidates. In this study, we investigate the usability of the graphics card to gain speed and accuracy in mapping the reads. Further, we compare the performance of different assembly programs taking false positives into account. Here, we defined a false positive rate based on the number of misplaced or missing nucleotides per mapped read. This not only gives an indication of the possible missing SNPs, but it also takes the edge effect of local alignment into account. We show that our approach is more accurate compared to currently used, and it achieves performance improvements of up to 112 fold, by using the graphics card.

Joint work: Fritz Sedlazeck 1;2;3;4, Greg Ewing 2;5 & Arndt von Haeseler1;2;3;4

1 Center for Integrative Bioinformatics Vienna (CIBIV), Max F. Perutz Laboratories, Vienna, Austria 2 University of Vienna, Vienna, Austria 3 Medical University of Vienna, Vienna, Austria 4 University of Veterinary Medicine, Vienna, Austria 5 Mathematics and BioSciences Group (MABS), Vienna, Austria