[modeller_usage] question about multiple alignment tutorial

[Mike White <>]

I have a question about the example script salign.py in the Advanced  
tutorial for version 8v2. I understand the logic of the script where  
salign is called three successive times incorporating more  
information each time in the form of the "weights" parameter, and one  
sees increase in alignment quality with each cycle. However, I am  
unclear of the purpose of the last call to salign after the alignment  
files are written. The tutorial text says that the last call is to  
generate a quality score, but after examining the log file the  
quality of the alignment after this call is worse than after the  
first three calls. I noticed that the rms_cutoff for the first three  
calls is 3.5, while that for the final call is 1.0 (presumably more  
stringent), and that the alignment type is "progressive" rather than  
"tree." Since the alignment files are already written, and this is  
what is used in the alignment with the sequenced to be modeled, why  
is this last step run?


As usual, thanks for all of your help

Mike White