Dear Bonvin lab
I have recently been docking several linear peptides (of around 20 residues) to a disordered protein target. I used the HADDOCK2.4 web server.
I noticed something quite interesting, but also confusing. Basically, I noticed that minor differences in input peptide conformation result in a significant difference in docking score, poses and clustering, even if the peptides have the exact same sequence. This was the case even when docking using identical parameters and receptor, and keeping the peptides as fully flexible each time. I keep also the target as fully flexible.
The differences in peptide conformations are due to whether I minimized them or not, or did a “montecarlo” simulation on them (all prior to docking). I was not sure what would be ideal..so that is why I tried different ways of preparing them.
For example, for one peptide, I received a docking score of -118. When docking the exact same peptide sequence but with a slightly different input conformation, I received a docking score of -29.
Usually, the peptides with a good score of <-100 do not form significant docking clusters (they are all individual conformations), which is unlike the peptides with higher (“worse”) docking scores. Although, the peptides with good docking scores (<-100) do usually show consistent and good docking scores across the first 10 conformations.
My questions are: Why does the input peptide conformation seem to matter so much when I keep my peptides as fully flexible regardless?
Any input on this is appreciated! 