Extensions of the PSMC model and more recent population structure

Richard Durbin
Sanger Institute

The Pairwise Sequentially Markovian Coalescent (PSMC) model structure was introduced by Li and Durbin (2011)
to fit a history of ancestral effective population size from a single diploid genome sequence by modelling the
joint distribution of lengths of and mutations in IBD segments (segments separated by ancestral recombinations)
between the two chromosomes. However, the resolution of the PSMC approach drops within the last 30 kyears
or so because there are relatively few segments coalescing within that time. In addition, the current model assumes
a uniformly mixing ancestral population, and is vulnerable to ancestral structure. Here I plan to discuss approaches
to address these two issues. Perhaps I will even have some results. I may also talk about how population structure
over more recent time scales (last few thousand years) affects imputation accuracy at low allele frequencies (1% and below), which is something of practical importance to large scale sequencing and human genetics programs.


Back to Workshop IV: Coancestry, Association, and Population Genomics