SIFTER Protein Function Prediction

We are awash in proteins discovered through high-throughput sequencing projects. As only a minuscule fraction of these have been experimentally characterized, computational methods are widely used for automated annotation. Unfortunately, these predictions have littered the databases with erroneous information, for a variety of reasons including the propagation of errors and the systematic flaws in BLAST and related methods. We have developed a statistical approach to predicting protein function that uses a protein family's phylogenetic tree, as the natural structure for representing protein relationships^1,2. We overlay on this all known protein functions in the family. We use a model of function evolution to then infer the functions of all other protein functions. Even our initial implementations of this method, called SIFTER (Statistical Inference of Function Through Evolutionary Relationships) have performed better than other methods in widespread use. SIFTER was recently honored as the best-performing sequence-based method in the Critical Assessment of Function Annotation ^3,4.