On the automatic annotation of gene functions using observational data and phylogenetic trees

Motivation: Gene function annotation is important for a variety of downstream analyses of genetic data. Yet experimental characterization of function remains costly and slow, making computational prediction an important endeavor. In this paper we use a probabilistic evolutionary model built upon phylogenetic trees and experimental Gene Ontology functional annotations that allows automated prediction of function for unannotated genes. Results: We have developed a computationally efficient model of evolution of gene annotations using phylogenies based on a Bayesian framework using Markov Chain Monte Carlo for parameter estimation. Unlike previous approaches, our method is able to estimate parameters over many different phylogenetic trees and functions. The resulting parameters agree with biological intuition, such as the increased probability of function change following gene duplication. The method performs well on leave-one-out validation, and we further validated some of the predictions in the experimental scientific literature. Availability: Our method has been implemented as an R package and it is available online at https://github.com/USCBiostats/aphylo. Code needed to reproduce the tables and figures can be found in https://github.com/USCbiostats/aphylo-simulations. Author summary: Understanding the individual role that genes play in life is a key issue in biomedical-sciences. While information regarding gene functions is continuously growing, the number of genes with unknown biological purpose is yet greater. Because of this, scientists have dedicated much of their time to build and design tools that automatically infer gene functions. In this paper, we present yet another attempt to do such. While very simple, our model of gene-function evolution has some key features that have the potential to generate an impact in the field: (a) compared to other methods, ours is highly-scalable, which means that it is possible to simultaneously analyze hundreds of what are known as gene-families, compromising thousands of genes, (b) supports our biological intuition as our model’s data-driven results coherently agree with what theory dictates regarding how gene-functions evolved, (c) notwithstanding its simplicity, the model’s prediction accuracy is comparable to other more complex alternatives, and (d) perhaps most importantly, it can be used to both support new annotations and to suggest areas in which existing annotations show inconsistencies that may indicate errors or controversies.