You are here

analyzing the protein protein docking without native structure

3 posts / 0 new
Last post
analyzing the protein protein docking without native structure


I have performed protein-protein docking  and I would like to do analysis on the  results. I do not  have a native  structure. I was wondering whether it is right to consider the lowest energy decoy as native and follow the steps explained in the Dr Meiler tutorial, such as plotting RMSD vs. Binding energy and expect it as funnel-like plot.  If you have any other suggestions in this regard, please let me know.

Thank you


Post Situation: 
Sun, 2020-10-11 20:42

Hi rohi,

It's good to see those tutorials are getting good use. Generating a funnel plot from the lowest energy structure is a commonly adopted strategy, as you have hit on already. I would also recommend to explore the conformational space through clustering and generate funnel plots for the lowest energy structures in the top clusters against the whole population. You might also calculate something like a Pnear value for each cluster. If you see multiple conformations with funnel plots then you should continue your modeling efforts by: a) generating more decoys (maybe up to 10,000); b) running relax on the top decoys in each cluster; c) adopting an ensemble docking approach after generating backbone diversity.

It's also always good to check the results against any known mutagenesis data. Is there a mutation that abrogates binding and the model shows this to be a critical contact? If yes, then your model might be the right one. If this critical site is far from the interface (and the abrogation cannot be explained for another reason, such as protein quality) then you might want to explore other conformations. Maybe do some local docking or ensemble docking starting from conformations that show this critical interaction. (You need to have confidence in the critical site when adopting this strategy, though, as mutations can alter binding for many reasons and data is often dirty.)


Tue, 2020-10-13 08:07

Thank you very much for the help.

Tue, 2020-10-13 09:15