Parts | Days | Selection | Search | Updates | Downloads | Help
DY: Fachverband Dynamik und Statistische Physik
DY 16: Statistical Physics in Biological Systems (joint session DY/ BP)
DY 16.1: Talk
Wednesday, April 2, 2014, 09:30–09:45, ZEU 160
Statistics of local multiple sequence alignments — •Pascal Fieth and Alexander K. Hartmann — Institute of Physics, University of Oldenburg
To assess the significance of alignment scores obtained by comparing DNA or amino acid sequences using sequence alignment, knowledge of the score distribution in the biologically relevant high-scoring region is necessary. The score distribution can analytically be shown to follow a Gumbel extreme value distribution for gapless local alignments. For gapped alignments, however, the distribution can only be obtained numerically. To cover the rare-event region of the distribution, studies of the score distribution of pairwise local alignments were done utilising parallel tempering[1]. They showed that, unlike predicted by previous simple sampling approaches, a Gaussian correction to the Gumbel distribution is necessary in case of finite sequence lengths. Here, this study is expanded to sum-of-pair scores of multiple sequence alignments, i.e. the alignments of more than two sequences, with gaps. Results will be shown for the score distributions of local multiple alignments and compared to previous results for global multiple alignments, where regions with probabilities smaller than 10−70 could be obtained.
[1] S. Wolfsheimer, B. Burghardt, A.K. Hartmann, Local sequence alignment statistics: deviations from Gumbel statistics in the rare-event tail, Algorithms for Molecular Biology (2007)