SIFTER
SIFTER, Statistical Inference of Function Through Evolutionary Relationships, "is a statistical approach to predicting protein function that uses a protein family's phylogenetic tree, as the natural structure for representing protein relationships." \cite{Sahraeian_2015} SIFTER was developed on the premise that sequence similarity does not necessarily imply phylogeny \cite{Engelhardt_2011} and may misinterpret branching and/or duplication events of the internal nodes of the phylogenetic model. The SIFTER algorithm consists of four steps: 1) Begin with building a phylogenetic tree of homologs for the target protein. 2) Add gene ontology annotations for all proteins on the tree. 3) Beginning with leaf nodes, populate annotations upward to the root using a statistical model of function evolution. This step takes into account branch lengths and speciation event duplication. 4) Take the complete set of gene ontology terms and populate the tree back down to the leaf nodes. 5) Calculate annotation term probabilities for all proteins in the family.