[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [modeller_usage] GA341/DOPE/MolPDF which one for model selection?



Yun He wrote:
Currently, I am working on a model based on a very low sequence-identity template (about 13%). I use an alignment to produce 30 models, and then I sort these models by their molpdf or GA341 or DOPE. However, the sorts turn out very different results, some models with higher GA341 scores takes higher DOPE scores. I am confused, which one should I choose?

No model assessment score is perfect, so you should not expect them to be perfectly correlated.

A 'good' model will have a GA341 score near 1.0 (larger is better) but a small DOPE score (DOPE scores, like molpdf, are "energies", so smaller or negative is better).

Neither DOPE nor GA341 show a lot of discrimination for poor models - e.g. you can't really say that a model with a GA341 score of 0.1 is 'worse' than one with 0.2 (the score was developed to tell between good and bad models, not to rank bad models). Anything worse than 0.6 or so is a bad model.

	Ben Webb, Modeller Caretaker
--
             http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage