Protein Molecular Function Prediction Based on the Phylogenetic Tree

We employ a novel method to construct a phylogenetic tree based on distance matrix among different protein molecular sequences, and present a statistical model to infer specific molecular function for unannotated protein sequences within the phylogenetic tree. Our method produced specific and consistent molecular function prediction across the P-falciparum family. For the P-falciparum family, it achieves 91.2% precision and 76.9% recall, outperforms the related method GOtcha and BLAST. Finally, we intend to improve our method through adopting a more appropriate feature extraction approach from the sequence or a better statistical inference model in the future.