Calibrating a coalescent simulation of human genome sequence variation

Abstract

Population genetic models play an important role in human genetic research, connecting empirical observations about sequence variation with hypotheses about underlying historical and biological causes. More specifically, models are used to compare empirical measures of sequence variation, linkage disequilibrium (LD), and selection to expectations under a "null" distribution. In the absence of detailed information about human demographic history, and about variation in mutation and recombination rates, simulations have of necessity used arbitrary models, usually simple ones. With the advent of large empirical data sets, it is now possible to calibrate population genetic models with genome-wide data, permitting for the first time the generation of data that are consistent with empirical data across a wide range of characteristics. We present here the first such calibrated model and show that, while still arbitrary, it successfully generates simulated data (for three populations) that closely resemble empirical data in allele frequency, linkage disequilibrium, and population differentiation. No assertion is made about the accuracy of the proposed historical and recombination model, but its ability to generate realistic data meets a long-standing need among geneticists. We anticipate that this model, for which software is publicly available, and others like it will have numerous applications in empirical studies of human genetics.The search for inherited influences on human disease and the effort to understand the history of human populations rely on a detailed knowledge of human genome sequence variation. Population genetic models serve as important tools in this quest. By simulating variation under neutral evolution, they provide background expectations about genetic variation, for example, for the frequency distribution of disease alleles and for patterns of linkage disequilibrium. They also serve as a tool to evaluate evidence for evolutionary selection, and to infer the history of populations.A primary difficulty in using such models lies in deciding which ones to employ. Given the complex history of human populations and the wide variety of plausible model parameters, as well as considerable uncertainty about how mutation and recombination rates vary, it is impossible rigorously to infer from data a single "correct" model: The problem is badly underdetermined. Until comparatively recently, in fact, genome-scale data were scarce enough that they provided few constraints on models. As a result, a common practice in simulating human variation has been to use a set of simple models that are easy to implement. The most basic model is the standard Wright-Fisher neutral model of a freely mixing, constant-sized population, with uniform rates of recombination and mutation across the genome. This model has been (and continues to be) widely used, as have other simple models (e.g., island models).Simple models have been of enormous utility as heuristics. In some respects, they have also offered surprisingly good fits to e...

Genome‐wide significance for dense SNP and resequencing data

“…The simulation was implemented using the FREGENE software [Hoggart et al, 2007], which simulates large genomic regions in large diploid populations, forward in time, under various demographic scenarios and evolutionary models, including variable recombination rates and gene conversion. In our simulation, we mimicked the modeling assumptions of [Schaffner et al, 2006] in which the demographic and evolutionary model of the simulation was chosen to approximate the history of populations from three continental regions: West Africa, East Asia and Europe. The genetic parameters of the simulation were tuned [Schaffner et al, 2006] to match (1) the allele frequency distribution; (2) the relationship between allele frequency and the probability that an allele is ancestral; (3) F st , and two measures of the extent of LD; (4) the relationship of genetic distance with r 2 and (5) the fraction of pairs of markers with D 0 5 1 in each of the three populations [Schaffner et al, 2006].…”

Sum of parts is greater than the whole: inference of common genetic history of populations

“…We use COSI [5] that is the only population simulator, to the best of our knowledge, that provides the ARG as well as produces populations that match the the genetic landscape of the observed human populations. We use the bestfit model in COSI to simulate the samples with a calibrated human demography for different populations, proposed by Schaffner et al [5].…”

“…We use the bestfit model in COSI to simulate the samples with a calibrated human demography for different populations, proposed by Schaffner et al [5]. This demography generates data matching three structured continental populations: Africans, Europeans and Asians.…”

“…However it has been seen that it can vary widely in a fine-scale manner when focusing on specific regions of the genome [12]. Different simulations are run using recombination rates matching the major portion of the range observed in human data [5]. …”

Positive selection on mitochondrial M7 lineages among the Gelong people in Hainan

“…To distinguish the effect of demographic expansion, we performed demographic coalescence simulations 10 000 times using a MS program. 15 Reference East Asian and African population models 16,17 were set as best-fit demographic models. The mutation rate was set as 3.24Â10 À6 per site per generation 18 (20 years per generation).…”