Hello everyone!
I know that I-TASSER can't use multiple restraints. However, I decided to try what will be if I will write several templates one after each other
in .fasta file (I provide the file here). And the result of such homology modelling was even better (I have performed quite an extensive benchmark already) than for pairwise modelling with any of the templates that were in this "multiple template restraint 2". From the log file it seems that I-TASSER used all these templates as a single concatenated template. But what was the mechanism of getting such a good result? What has happened and why did it finally give me such a good result rather than nonsense?