I have been docking a cyclic peptide to its receptor for a while now. The affinity comes back around -9 kcal/mol no matter if I change the chirality of the peptide's amino acids, or if I increase the exhaustiveness of the search.
In trying to assess whether this binding figure was a good representation of the peptide's actual affinity, I went ahead and removed all side chains from the peptide to yield what is basically a cyclic polyglycine. I assumed that the affinity would be either zero or some positive number. To my surprise, however, the affinity came back as -7 kcal/mol. This leads me to believe that the docking of my actual peptide is probably not good.
Is there a way to look at this data and have some level of confidence that the docking was successful? is there a threshold people commonly use to determine if the docking was good? I've found a tutorial on how to dock cyclic molecules using Autodock 4.2., but could not find anything for Autodock Vina. I am not sure if this means that the latter can not do it, or if it can do it without the need for any extra steps (like breaking a bond).
Any feedback would be greatly appreciated.
Thanks.
Abdullahi Ibrahim Uba thank you so much for your answer. I’ll look into another platform to complement autodock. Any suggestions?
Dear Federico,
You have asked unexpectedly deep question. You really asked not how to interpret a particular result but how good any theoretical methods in structural biology are. The general answer is: they are weak. But we have no other tools so we use them. (Exactly like a rigged roulette table that is only one in town). You have to develop (by theoretical experimentation with different programs) an intuitive set of tools that helps you to interpret or trust a particular program depending on the particular situation. I do not have the time and there is no place to list all the weaknesses. However there is a few rules of thumb.
1) Seek consensus on the pose nad the numerical output against a whole range of programs. There are consensus programs available also.
2) Check with an independent method such as placing the best solution and run a long MD. MD suffers from the same weaknesses but the applications are different and the benchmarks easier to comprehend.
3) If you really starving for basic science use more sophisticated methods such as QM ab initio or functional density methods.
4) Use independent experimentation to gauge the results from individual program.
5) Use advice from experienced people.
As a guiding principle you can adopt what we have adopted in an obscure but reasonable paper: Shape Matters: Improving Docking Results by Prior Analysis of Geometric Attributes of Binding Sites. RM Keyes, E Pejo, K Katagiri, K Huynh, A Rudnitskaya, B Stec, KA Stieglitz.
JSM Chemistry 4 (2016) 1020
Good Luck.
Bog
Thank you all for your advice. It has been really helpful.
Best regards.
- Badges
- Science topic
- Similar topics
- Microbiology
- Assays