The Discovery of Single-Nucleotide Polymorphisms—and Inferences about Human Demographic History

摘要: A method of historical inference that accounts for ascertainment bias is developed and applied to single-nucleotide polymorphism (SNP) data in humans. The consist 84 short fragments the genome were selected, from three recent SNP surveys, contain at least two polymorphisms their respective samples then fully resequenced 47 globally distributed individuals. Ascertainment deviation, what would be observed a random sample, caused either by discovery small or locus selection based on levels patterns polymorphism. surveys which present derived differ both protocols size used discovery. We implemented Monte Carlo maximum-likelihood fit subdivided-population model includes possible change effective some time past. Incorrectly assuming does not exist causes errors inference, affecting estimates migration rates changes size. Migration are overestimated when ignored. However, direction error inferences about population (whether inferred shrinking growing) depends whether numbers SNPs per fragment SNP-allele frequencies analyzed. use abbreviation “SDL,” “SNP-discovered locus,” recognition genomic-discovery context SNPs. When modeled fully, number SDL allele support scenario growth subdivided population. If subdivision ignored, however, hypothesis constant cannot rejected. An important conclusion this work that, demographic other studies, useful only extent can modeled.