The massive accumulation of protein sequences arising from the rapid development of high-throughput sequencing, coupled with automatic annotation, results in high levels of incorrect annotations. In this study, we describe an approach to decrease annotation errors of protein families characterized by low overall sequence similarity. |