Refinement of Protein Structures with HADDOCK


Using AlphaFold2 I built a model of a homodimeric protein that has a total of 3128 amino acid residues. 83.2% of its residues are in the most favored regions, 13.7% are in the additional allowed regions and 3.1% are in the generously allowed regions. I want to refine this model with HADDOCK2.4, but I have a few doubts. The first is which protocol to use. The choice depends on what? Should I use all the protocols? Is there a protocol that is the most widely used, for example “Water Refinement”? As for “Clustering”, should I leave “Method”, “Cutoff” and “Minimum cluster size” at FCC, 0.60 and 1 respectively? Or should I change it to RMSD, 2.0, 1? In “Analysis parameters”, should I leave “Cluster” or change it to “Full”? I looked for this information in the tutorials provided, but couldn’t find it. I haven’t found how to interpret the results either.

Another question I have does not concern HADDOCK2.4, but the refinement itself. According to your experience, should refinement be carried out until 98% residue is reached in the most favorable regions of the Ramachandran graph, or is refining just once enough since refining doesn’t always improve anything?

In this case I would use the standard water refinement protocol. Assuming you don’t have too many clashes.

About the most favourable Ramachandra percentages, you should not expect to reach 98% … This is not what you observe in experimental structures… And longer refinement protocols will not per se improve those percentages

More reading about refinement protocols and their performance:

1 Like

Copy that! You helped me a lot, thank you very much!