Multipoint likelihoods for genetic linkage: The untyped founder problem
- Univ. of Pittsburgh, PA (United States)
Too many untyped founders in a pedigree cause the Elston-Stewart algorithm to grind to a halt. Our solution to this problem involves recoding alleles based on symmetry and identity-by-descent to greatly reduce the number of multi-locus genotypes. We also use modified genotype elimination to better organize the calculation, substantially reducing the amount of memory needed. We never have to consider multi-locus genotypes that are not valid. Thus for typed pedigrees, the calculation is independent of the number of alleles at a locus. In addition, our locus-by-locus method allows us to group similar calculations to avoid recomputation, costly bookkeeping for valid genotypes, and large memory allocation. We were able to do a 4-locus likelihood for a 41-member simple pedigree with the first two generations untyped and an allele product of over 1500 in under an hour. This likelihood cannot be computed at all with LINKAGE, since some of its arrays would require over a gigabyte of memory. Our locus-by-locus method is also well-suited for parallelization since we can factor the computation into smaller independent pieces. This will enable us to tackle problems of even greater complexity.
- OSTI ID:
- 134709
- Report Number(s):
- CONF-941009-; ISSN 0002-9297; CNN: NIH grant HG00932; TRN: 95:005313-1447
- Journal Information:
- American Journal of Human Genetics, Vol. 55, Issue Suppl.3; Conference: 44. annual meeting of the American Society of Human Genetics, Montreal (Canada), 18-22 Oct 1994; Other Information: PBD: Sep 1994
- Country of Publication:
- United States
- Language:
- English
Similar Records
Monte Carlo multipoint linkage analysis
Incremental computation of multipoint likelihoods in constructing genetic linkage maps