Genetics · 2004 · 388 citations · 32 references
Public HealthBiostatisticsGeneticsGenomic Data AnalysisPopulation GeneticsStatistical GeneticsPedigree Information IntegrationPopulation Genetics InferenceLikelihood MaximizationMaximum Likelihood EstimatorMethod Of Maximum LikelihoodSibship AnalysisType I Error Rate
Likelihood methods partition individuals into full‑ and half‑sib families from genetic markers, but they assume error‑free data, an assumption rarely met that can bias sibship estimates. This study proposes a new likelihood approach that incorporates simple, robust models of typing error. The method iteratively updates allele frequencies using reconstructed sibships, allows parental information, and employs efficient algorithms for likelihood calculation and optimization, and it has been extensively tested on simulated and empirical datasets. Simulations demonstrate that the method accurately infers full‑ and half‑sibships even with high error rates and can pinpoint typing errors at each locus within each reconstructed family.
Abstract Likelihood methods have been developed to partition individuals in a sample into full-sib and half-sib families using genetic marker data without parental information. They invariably make the critical assumption that marker data are free of genotyping errors and mutations and are thus completely reliable in inferring sibships. Unfortunately, however, this assumption is rarely tenable for virtually all kinds of genetic markers in practical use and, if violated, can severely bias sibship estimates as shown by simulations in this article. I propose a new likelihood method with simple and robust models of typing error incorporated into it. Simulations show that the new method can be used to infer full- and half-sibships accurately from marker data with a high error rate and to identify typing errors at each locus in each reconstructed sib family. The new method also improves previous ones by adopting a fresh iterative procedure for updating allele frequencies with reconstructed sibships taken into account, by allowing for the use of parental information, and by using efficient algorithms for calculating the likelihood function and searching for the maximum-likelihood configuration. It is tested extensively on simulated data with a varying number of marker loci, different rates of typing errors, and various sample sizes and family structures and applied to two empirical data sets to demonstrate its usefulness.
32
Optimization by Simulated Annealing
Scott Kirkpatrick, C. D. Gelatt, M.P. Vecchi · Science · 1983 · 44.6K citations
Statistical confidence for likelihood‐based paternity inference in natural populations
Tristan Marshall, Jon Slate, Loeske E. B. Kruuk et al. · Molecular Ecology · 1998 · 4.3K citations
Life Sciences, Biology, Genetics +10
Genetics and analysis of quantitative traits
James M. Cheverud · American Journal of Physical Anthropology · 1999 · 3.9K citations
Life Sciences, Biology, Genetics +12
Reliable genotyping of samples with very low DNA quantities using PCR
Pierre Taberlet · Nucleic Acids Research · 1996 · 1.5K citations · Full text
Life Sciences, Biology, Genetics +13