Hello,
I'm working on a project of a proteome screening by modelisation and function prediction with the stand-alone I-TASSER 3.0 and I want to know how can I estimate how significant are the functions predicted with the different scores associates.
In the case of the output file obtained for the GO terms with COFACTOR : GOsearchresult_model1.dat file. There is a GO-score but there is no indication of the range for this score. In you're publication (I-TASSER: a unified platform for automated protein structure and function prediction) you've said that "using a GO-score cutoff of 0.5, 85.1% of molecular functions, 76.9% of biological processes, and 74.6% of cellular locations can be correctly assigned".
How confident can I be with lower cutoff, like 0.45 or 0.35 ?
For the EC predictions, in the same publication you mention "using an EC-score cutoff of 1.1, the first three EC digits can be correctly assigned in 72.4% cases". In the README.txt file you mention an FC-score for the stand-alone version and there is no EC-score calculated. How interpret this FC-score ?
In the same way I don't know how to interpret the score (BS-score) obtained for the Ligand-binding sites predictions. It would be greatly appreciated if you indicate us the range of this score and any informations/cutoff to get confidence in the predictions ?
Thank you,
Francois Inizan