PSEUDOMARKER 2.0: efficient computation of likelihoods using NOMAD

Friday, November 08, 2013 — Poster Session IV
2:00 p.m. – 4:00 p.m.	FAES Academic Center (Upper-Level Terrace)	NLM	COMPBIO-16

Authors

A.A. Schaffer
E.M. Gertz
T. Hiekkalinna
S. Le Digabel
C. Audet
J.D. Terwilliger

Abstract

PSEUDOMARKER is a software package that tests for significant relationships between a putative trait-predisposing locus and genotypes at markers, such as microsatellites or SNPs. Typically, such packages either test for genetic linkage or test for linkage disequilibrium (LD). With PSEUDOMARKER, one can jointly analyze linkage and LD, or LD conditional on linkage, using pedigree relationships and singleton individuals. PSEUDOMARKER version 1 maximizes several likelihood functions using a generalized pattern search (GPS) algorithm implemented in a custom version of the ILINK program. The running time of PSEUDOMARKER depends on the number of times the optimization algorithm evaluates a likelihood function. We modified ILINK to use the software NOMAD to maximize likelihoods instead of GPS, resulting in generally comparable or better optima with many fewer evaluations of the likelihood functions. Out of 288 tests, NOMAD reported a log likelihood more than 0.005 better than that found by GPS 68 times, while GPS was at least 0.005 better only 7 times. On 12 datasets with multiple tests each, the number of likelihood evaluations in PSEUDOMARKER 2.0 is reduced by a factor of at least 2 on each dataset and by more than 7 on the best dataset. PSEUDOMARKER is available from http://www.helsinki.fi/~tsjuntun/pseudomarker.