SSEalign: accurate function prediction of bacterial unannotated protein, based on effective training dataset

The functions of numerous bacterial proteins remain unknown because of the variety of their sequences. The performances of existing prediction methods are highly weak toward these proteins, leading to the annotation of \"hypothetical protein\" deposited in NCBI database. Elucidating the functions of these unannotated proteins is an urgent task in computational biology. We report a method about secondary structure element alignment called SSEalign based on an effective training dataset extracting from 20 well-studied bacterial genomes. The experimentally validated same genes in different species were selected as training positives, while different genes in different species were selected as training negatives. Moreover, SSEalign used a set of well-defined basic alignment elements with the backtracking line search algorithm to derive the best parameters for accurate prediction. Experimental results showed that SSEalign achieved 91.2% test accuracy, better than existing prediction methods. SSEalign was subsequently applied to identify the functions of those unannotated proteins in the latest published minimal bacteria genome JCVI-syn3.0. Results indicated that At least 99 proteins out of 149 unannotated proteins in the JCVI-syn3.0 genome could be annotated by SSEalign. In conclusion, our method is effective for the identification of protein homology and the annotation of uncharacterized proteins in the genome.