SEARCH FOR OVERREPRESENTED CHARACTERISTICS OF GENES: IMPLEMENTATION OF PERMUTATION TESTS USING GPUs
A. A. Yakimenko1, K. V. Gunbin2, M. S. Khairetdinov3
1Novosibirsk State Technical University, pr. Karla Marksa 20, Novosibirsk, 630073 Russia
2Institute of Cytology and Genetics, Siberian Branch, Russian Academy of Sciences, pr. Akademika Lavrentieva 10, Novosibirsk, 630090 Russia
3Institute of Computational Mathematics and Mathematical Geophysics, Siberian Branch, Russian Academy of Sciences, pr. Akademika Lavrentieva 6, Novosibirsk, 630090 Russia
Keywords: GPUs, matrix-vector operations, permutation test
Abstract
An algorithmic approach and software for solving the problem of search for statistically significant overrepresented biological characteristics of genes from a given set are proposed. The problem is related to implementation of permutation (randomization) tests well known in biology. Parallel software based on the potential parallelization of the permutation test is developed and implemented on GPUs. The efficiency of application of this software on empirical materials is estimated.
|
