ClassX: a browsing tool for protein sequence megaclassifications

The authors have developed an algorithm, HHS, for efficient clustering of very large sequence databases into groups based on sequence similarity. When applied to protein sequence databases, the induced groups objectively define both protein superfamilies and recurring sequence motifs. HHS has been successfully applied to large sequence databases; however, the output is too large and complex to be readily interpreted. The authors report on a software tool called ClassX that was developed to help explore the results of megaclassification runs. They also discuss the problem of automatic annotations, i.e., how to use the classifications produced by the program to label regions of interest in an anonymous protein sequence.