An improvement method based on OrthoMCL by adding the domain information

Existing clustering methods have so far not separated paralogs from orthologs effectively. Since proteins evolve with their structural and functional domains as independent units, to achieve a higher level of sensitivity and specificity in assessing the similarity, it is necessary to add the domain information. We presented a method to improve the clustering results of the orthologs and paralogs from multiple species by adding the domain information. First, we do the all-against-all blast between the protein sequences. At the same time, we make the blast between the sequences and Pfam-A database to find which of the sequences share the same domain. Then we use this information as an additional criterion for filtering false relationships in all-against-all BLASTP results, and generate a similarity matrix. Final, the MCL algorithm is applied to group orthologs from multiple species. Our preliminary results show that our method can improve strikingly the precision of sequence clustering.