Hi Aron
We usually base on analysis on the HADDOCK score.
Relevant points to consider:
- is the score of the best cluster significantly different from the second best and such?
- how many models to cluster? I.e. is there some convergence in your simulations?
- how do the plots look like? E.g. a “potato” of points is not indicative of good convergence.
- Do you see convergence / “binding funnels”?
- Do your model satisfy the input data?
- …
Check our various online tutorials at Education & Tutorials of the Bonvin Lab – Bonvin Lab
And also our best practice guide at https://www.bonvinlab.org/software/bpg/
and in particular: https://www.bonvinlab.org/software/bpg/analysis/#dos-and-donts