Multipe Sequence Alignments:
Using the T-Coffee Web Server
Solution




  1. In the Default T-Coffee alignment, the red portions are the most likely to be correct. You can compare this alignment with the reference alignment as provided in prefab (use the command line t_coffee -other_pg aln_compare -al1 aln1 -al2 aln2 if you want a more precise comparison). For your information, 27 % of the positions are correct in this alignment when considering the two structures 1svpA and 1qq4A.


  2. Expresso uses BLAST to find the best structural templates, and reports them in the template_file. Expresso then combines the structure based alignments (sap) with the sequence based alignment methods (slow_pair, lalign_id_pair). Overall, the alignment is better, but this does not show very well in the scoring scheme (*) because the sequence based alignments (slow_pair, lalign_id_pair) are still as unconsistent (and unreliable) as they were before. For your information, 69% of the positions are correct in the alignment of 1svpA with 1qq4A

  3. M-Coffee produces a multiple sequence alignment with the most common methods. Because they tend to use similar algorithm, these methods often produce more consistent output than the default T-Coffee, thus making the consistency output slightly less informative. However, you can safely assume that highly consistent regions (red brick columns) are correctly aligned (*). In this case, the accuracy for the two structures is about the same as the default T-Coffee (27%).

  4. You can run Expresso without using the sequence based methods by going into the advanced mode and unchecking the slow_pair and lalign_id_pair boxes and run your job. As you can see (*) the consistency score is now much higher. This is because the pairwise structural alignment tend to agree very much with one another. This alignment is a whooping 91% correct with respect to the two considered structures. The sequences not colored are those for which a structure was not available.

Questions should be sent to C.Notredame