we are getting mad trying to understand how to get the value for the scorefilter option for a global search from a calibration run.
We have generated 5000 decoys with no filters. Then ? We simply got lost here, and we also feel we are missing some concepts…
Is the scorefilter applied as the coarse grained decoy should be passed to the full atoms section ? That is, it is a scoring of the coarse grained decoy ? As far as we understood, all the values in the fasc file as well as in the generated decoys refer to the full atom structure, so they are not useful to calibrate the scorefile value. Are we wrong ?
Should we run rosetta in score mode on the 5000 decoys to get this ? Final question, supposing we have a list of all the values, how to choose the scorefilter value ? +X of the best score ? In the case, how big X should be ?