PAG-I Plant Genome I Conference

Town & Country Conference Center, San Diego, CA, November, 1992.


PG-I: 93pg1

LOCUS ORDERING USING ISOLATED ANNEALING

Ben Hui Liu and Steve J. Knapp, Department of Crop and Soil Science Oregon State University, Corvallis, OR 97331


Locus ordering is the most computationally intense phase of genetic mapping. The computational burden increases exponentially as the number of linked loci increases. For k linked loci there are k!/2 testable locus orders. When k is big it is impossible to evaluate the testable orders even with a supercomputer. Simulated annealing algorithm (SA), which is an analogy of thermodynamics, is a technique suitable for optimization of large scale problems and has been used for designing complex integrated circuits and for business management. Here we report a simulation study on SA for locus ordering using the sum of adjacent recombination frequencies (SAR) as objective function. Factorial combinations of the following were simulated for 1000 replications: population size (n) of 50, 100, and 200; number of loci in a linkage group (k) ( 8, 30, and 100); and locus spacing pattern (in cM) of 1,1, 1,...,5,5,5,...,I,5,1,5,...,20,20,20,..., and 5,20,5,20,.... For 8 locus linkage roups, all possible orders were evaluated directly to find the most probable order (DIRECT) in order to compare to the order found by SA. The simulation results between DIRECT and SA were always the same or very close in terms of percentages of correctly estimated locus orders (PCO) and mean SARs. For locus spacing patterns 1,1,1,... and 1,5,1,5,... PCOs were low for all combinations of n and k. For the other locus spacing patterns, PCOs were generally higher than 90% when n = 100 and 200. We conclude that SA is efficient for locus ordering. The size of the mapping population should be sufficiently large in order to resolve locus order for dense linkage. SA and resampling techniques can be combined to estimate locus order. Locus ordering using SA was implemented in GMENDEL, a program we developed for genetic mapping.


Return to Previous Page or Intl-PAG Homepage