Send to

Choose Destination
Bioinformatics. 2004 Apr 12;20(6):818-28. Epub 2004 Jan 29.

Quality of alignment comparison by COMPASS improves with inclusion of diverse confident homologs.

Author information

Howard Hughes Medical Institute, and Department of Biochemistry, University of Texas Southwestern Medical Center, 5323, Harry Hines Blvd, Dallas, TX 75390-9050, USA.



Adding more distant homologs to a multiple alignment and thus increasing its diversity may eventually deteriorate the numerical profile constructed from this alignment. Here, we addressed the question whether such a diversity limit can be reached in the alignments of confident homologs found by PSI-BLAST, and we analyzed the dependence of the quality of the profile-profile comparison made by COMPASS on the sequence diversity within these alignments.


Protein families that have a greater number of diverse confident homologs in the current sequence databases provide an increased quality of similarity detection in profile databases, but produce on average less accurate profile-profile alignments with their remote relatives. This lower alignment accuracy cannot be improved when the most distant members of these families are excluded from their profiles. On the contrary, the presence of more diverse members results in more accurate alignments. For families with a high diversity of confident homologs, the lower quality of profile alignments with their remote relatives seems to be an attribute of these families or their alignments, rather than to be caused by the large number of diverse sequences itself. Our results suggest that at any level of profile diversity, one should include in the multiple alignment as many confident sequence homologs as possible in order to produce the most accurate results.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Silverchair Information Systems
Loading ...
Support Center