Novel metrics for evaluating the functional coherence of protein groups via protein semantic network.

Department of Biostatistics, Bioinformatics and Epidemiology, 135 Cannon Street, Charleston, South Carolina 29425, USA.

We present the metrics for assessing overall functional coherence of a group of proteins based on associated biomedical literature. A probabilistic topic model is applied to extract biologic concepts from a corpus of protein-related biomedical literature. Bipartite protein semantic networks are constructed, so that the functional coherence of a protein group can be evaluated with metrics that measure the closeness and strength of connectivity of the proteins in the network.

PMID: 17672896 [PubMed - indexed for MEDLINE]

PMCID: PMC2323239