AHRD: Automated Assignment of Human Readable Descriptions and Gene Ontology terms

We present a tool that assigns human readable descriptions (HRD) and Gene Ontology (GO) terms to unknown protein sequences, thus proving protein function prediction. AHRD scores BLAST hits taken from searches against different databases on the basis of the trust put into these databases and the local alignment quality. The BLAST hit descriptions are tokenized into informative words and a lexical analysis scores these tokens according to their frequency and the quality of the BLAST hits they occur in. Finally the best scoring description is assigned, and Gene Ontology terms transferred.

We have evaluated the performance of AHRD on different test sets and optimized parameters. For HRD, an F1 score is computed based on words shared with the reference annotation. For GO terms, we implement both a simple scoring based on shared GO terms considering all ancestor terms as well as semantic similarity. AHRD outperforms Blast2GO, Interpro2GO and best blast based methods.