Prediction of protein function by combining phylogenetic tree and mathematical inference

Sequences lacking molecular function annotations have limited research value, so automated approaches for annotating molecular function are used to smooth over this deficiency. Whereas massive erroneous prediction results that may reduce the reliability and utility of these methods is non-ignorable in protein function prediction. Approaches combined phylogenomics and mathematical inference have been developed to label the proteins. However, real phylogenetic tree is always unknown and the prediction accuracies of the existing approaches need to be increased in practical applications. In order to take advantage of evolutionary information extracted from the protein sequences, we construct a phylogenetic tree based on the matrix that indicates the distance among different protein molecular sequences. Then an inference model was used to infer the specific molecular function of un-annotated proteins in phylogenetic tree. It generates more satisfactory results than the previous methods. The results show that it is a promising approach to predict the molecular function of proteins.