On Dec 10, 2007, at 2:54 PM, Eric Pettersen wrote:
In regards to Match->Align producing a different number of aligned columns than the MatchMaker "core" even with the same cutoff value, that is totally believable. You will get situations where loop residues happen to cross each other within the cutoff distance but those residues were not in the same column of the MatchMaker alignment [nor should they be really] and therefore could not be in it's final "core".
Just wanted to add: I've found that with more difficult-to-align sequences (worst case is very low sequence identity and all-beta or all-alpha secondary structure), there may be incorrect segments in the initial MatchMaker alignment that are corrected in alignment from a subsequent Match-
Align step. The purpose of MatchMaker is to generate a correct superposition, and this is still successful in nearly all cases because the incorrect areas are pruned during fit iteration (only the correct columns are used to generate the final superposition). Match- Align will then generate columns for all the superimposed parts, some of which were not used in the prior fitting step.
Which alignment is better depends on the situation. For example, there could be different structures of the same protein where one loop moves a lot. The MatchMaker alignment will simply align the entire identical or nearly identical sequences, whereas Match->Align will not align the loops that are poorly superimposed in space. If you are comparing different structural matches, especially hard-to- align distantly related cases, it may be more appropriate to use the number of pairs in the Match->Align alignment rather than those values from MatchMaker. In our paper, for example, we reported the number of pairs and RMSDs from the Match->Align alignment, not the MatchMaker one: Tools for integrated sequence-structure analysis with UCSF Chimera: E.C. Meng, E.F. Pettersen, G.S. Couch, C.C. Huang, and T.E. Ferrin, BMC Bioinformatics 7, 339 (2006). http://www.biomedcentral.com/ 1471-2105/7/339 The drawback is that getting that RMSD value requires another round of fitting, this time on all positions in the Match->Align alignment (without iteration). This may have been more detail than you wanted! Elaine ----- Elaine C. Meng, Ph.D. meng@cgl.ucsf.edu UCSF Computer Graphics Lab and Babbitt Lab Department of Pharmaceutical Chemistry University of California, San Francisco http://www.cgl.ucsf.edu/home/meng/index.html