| Background and GoalDuring the expansion of human ancestors from Africa to the whole world,in order to adapt to various environmental factors,including climate,pathogens,land farming conditions,etc.,different populations experienced a large number of adaptive evolution events and produced different favorable mutations in their genomes.Adaptive evolution is associated with susceptibility to many diseases and response to many drugs,because an "adaptation" and "advantage" tens or thousands of years ago may no longer be an advantage given today’s environment and lifestyle.At present,there are still few favorable mutations known in the genomes of different human populations,because it is very difficult to identify favorable mutations of a single base,and the existing methods for identifying favorable mutations have obvious shortcomings.The purpose of this study is to develop a reliable method for identifying favorable mutations,and analyze the association of favored mutations in different populations with disease susceptibility in human populations,revealing that these associations are of great significance for precision medicine at the population and individual levels.Method and ResultFirstly,this study developed a method identifying favored mutations called DeepFavored.The key designs of DeepFavored include:splitting neutral mutations into hitchhiking mutations and ordinary mutations,thus splitting the task identifying favored mutations into two sub tasks,i.e.distinguishing favored mutations from hitchhiking mutations and distinguishing favored mutations from ordinary mutations;developing a deep learning based model to solve the two sub tasks simultaneously,and integrating the outcomes of the two sub tasks to identify favored mutations.Using simulated and real data,taking diverse scenarios of adaptive evolution into account,and using multiple statistics measuring method performance,this study comprehensively evaluated the performance of DeepFavored.The performance evaluation indicates DeepFavored outperforms the state of art competing methods.Secondly,this study used DeepFavored to massively identify the favored and hitchhiking mutations in European,Asian and African populations,and conducted joint analysis on the identified mutations,functional gene sets and disease-associated sites.The results of the joint analysis include:favored mutations and disease-associated sites are co-enriched on specific gene sets;favored and hitchhiking mutations are enriched on disease-associated sites;and the strengths of positive and negative correlations between favored mutations and disease mutations are similar.These results indicate:there an extensive trade-off between adaptive evolution and disease susceptibility,and the two controversial scenarios,i.e adaptive evolution increase disease susceptibility and disease susceptibility interferes with adaptive evolution,co-exist widely.Finally,this study used DeepFavored to identify whole-genome wide favored and hitchhiking mutations of 17 European,Asian and African populations,and developed a database with flexible search engine and genome browser to integrate the identified favored and hitchhiking mutations and multiple types of data,enabling researchers to dive into the association between adaptive evolution,disease susceptibility,drug response and the expression and function of genes.Summary and DiscussionThis study developed a more powerful method identifying favored mutations,which has great significance to examine the history of human adaptive evolution,the molecular mechanisms of important human traits.This study discovered that there an extensive trade-off between adaptive evolution and disease susceptibility,which can provide interpretations of the evolution of pathogenic mutations and the disease susceptibility differences between human populations.The database built here promises to promote the researches of the association between adaptive evolution,disease susceptibility,drug response and the expression and function of genes,providing great significance to precision medicine. |
