Genetic History of Populations: Limits to Inference

Abstract

AbstractThe dispersal of the human population to all the continents of the globe is a compelling story that can possibly be unravelled from the genetic landscape of the current populations. Indeed, a grasp on this strengthens the understanding of relationship between populations for anthropological as well as medical applications. While the collective genomes is believed to have captured its own “evolution”, undoubtedly, a lot is irrecoverably lost. It is important to try to estimate what fraction of past events become unreconstructable. We used a published population simulation algorithm which includes parameter sets that simulate modern regional human demographics: it reflects the out-of-Africa expansion events, isolation during the Last Glacial Period, Neolithic expansion of agriculture, and the industrial revolution for African, African–American, European, and Asian populations. This simulation tool was used to provide complete genetic histories unavailable in real human population data. Next we used the minimal descriptor device which is an algorithm-independent means to explore the (potential) recoverable genetic events from the final extant population data. We found that, on average, around 65 % of the total number of genetic events are recoverable, with substantial variations among histories. We also found that increases of sequence length tended to yield diminishing returns in new information yield. Lastly, even with a substantial fraction of events unrecoverable, and even for different population history simulations, the recoverable events do yield similar resolution of the whole record of demographic, climatic, and other population events.KeywordsRecombination RateOutgoing EdgeGenetic LandscapeSpecific HistoryGenetic HistoryThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.