On 02/29/2012 05:28 AM, mircial@sjtu.edu.cn wrote: > I am a new user of the MODELLER package. I am using the modeller > package to refine a loop region of a X-ray crystal structure. I have > finished modelling 100 independent structures of that loop and in > order to pick out the "good" structures, I need to do cluster > analysis on these structures and then compare the representative > models from each cluster, am I right?
That would be one way to tackle the problem, yes, but there are various ways you could do it (the ModLoop web server, for example, simply picks the one model with the lowest energy; often visual inspection of an ensemble is sufficient to identify the largest cluster).
> My question is, could MODELLER package do this cluster analysis on > the loop structures I got?
While Modeller does include some basic clustering methods, you are probably better off playing with specialist software for this to find your favorite method.
Ben Webb, Modeller Caretaker