Predictions of Native American Population Structure Using Linguistic Covariates in a Hidden Regression Framework

摘要: BACKGROUND: The mainland of the Americas is home to a remarkable diversity languages, and relationships between genes languages have attracted considerable attention in past. Here we investigate which extent geography can predict genetic structure Native American populations. METHODOLOGY/PRINCIPAL FINDINGS: Our approach based on Bayesian latent cluster regression model membership explained by geographic linguistic covariates. After correcting for effects, find that inclusion information improves prediction individual clusters. We further compare predictive power Greenberg's Ethnologue classifications Amerindian languages. report classification provides better proxy than at stock group levels. Although high values be achieved from classification, nevertheless emphasize Choco, Chibchan Tupi families do not exhibit univocal correspondence with CONCLUSIONS/SIGNIFICANCE: class described here efficient predicting population using