Entering edit mode
11.3 years ago
Dollas Salleh
▴
70
In term of Homology Modelling Prediction. What criteria that must be prioritized in order to choose the best template from PDB. Whether E-value, identity , Gaps, query etc??
Thanks
In my experience, the best way is to look at the existing literature on the target protein family i.e. Kinases to find out the conserved motifs if there is any. Then the alignment obtained from Clustal or tcoffee or muscle should be modified manually. If the homology is low, then threading approaches (HHPred) or Rosetta can be more useful. All I said is true for soluble proteins. It all depends on the research question to address.
2 things that bear in my mind 1)Ive try to blast my target protein with PDB. The highest similarity up to 28% but different e value. Based on blast let say 28% identity with 2e-16 e value, 26% identity with 2e-18 e value and 36% identity with 1e-6 e value. Based on your opinion what criteria that must be prioritized whether e value or similarity? 2) I alredy try HHPred. Ok. Actually the result give a list of template according to e-value etc. But, my problem is how to choose suitable template from the result by HHpred??
Thanks. Sory if this question look like silly for you. And sorry for my bad english.
It is hard to say without looking at the alignments. I wonder what is the aim of the homology modeling. A homology model must be validated by biochemical experiments or comparing with existing literature i.e. cysteine accesibility studies (SCAM) if you want to publish them. Alternatively I-Tasser makes quite good models completely automatically. Your supervisor should assist you in it.