[Chimera-users] representative fits in chimera fitmap search

Jan Kosinski kosinski at embl.de
Fri Mar 18 10:19:59 PDT 2016


In the fitmap command global search, how the unique fit is selected from 
the cluster of similar fits?

Ideally, from each cluster, I would like to obtain the fit the gave the 
best cross-correlation. But, I think, based on the fit_search code in 
FitMap/search.py, what is happening is that the first unique fit ever 
encountered is added to the fit list:
         close = b.close_transforms(ptf)
         if len(close) == 0:
             transforms = [M.multiply_matrices(ptf, mtv) for mtv in 
             stats['hits'] = 1
             f = Fit(models, transforms, volume, stats)
             f.ptf = ptf
             fo[id(ptf)] = f
             s = fo[id(close[0])].stats
             s['hits'] += 1
and any subsequent fit that is close to the first, would be discarded, 
even if it gives better cross-correlation.

Do I understand correctly that the unique fit is not necessarily the 
best scoring fit from the cluster?

Thanks in advance for clarification,

More information about the Chimera-users mailing list