Computational methods for physical mapping of chromosomes
1991
Torney, D.C. | Schenk, K.R. | Whittaker, C.C. | White, S.W.
A standard technique for mapping a chromosome is to randomly select pieces (with replacement), to use restriction enzymes to cut these pieces into (sequence specific) fragments, and then to use the fragments for estimating the probability of overlap of these pieces. (Overlapping pieces are likely to 'share' fragments). Typically, the order of the fragments within a piece is not determined, and the observed fragment data from each pair of pieces must be permuted N1! X N2! ways to evaluate the probability of overlap, N1 and N2 being the observed number of fragments in the two selected pieces. We will describe computational approaches used to substantially reduce the computational complexity of the calculation of overlap probability from fragment data. Presently, about 10(-4) Cpu seconds on one processor of an IBM 3090 is required for calculation of overlap probability from the fragment data of two randomly selected pieces, with an average of ten fragments per piece. A parallel version has been written using IBM clustered FORTRAN. Parallel measurements for 1, 6, and 12 processors are presented. This approach has proven promising in the mapping of human chromosome 16 at Los Alamos National Laboratory.
显示更多 [+] 显示较少 [-]