mpDockQ/pDockQ scores seem >1 for 3-protein complexes about alphapulldown HOT 6 CLOSED

Thank you for your help. I will mark this as closed.

from alphapulldown.

Hi @niradsp

Thanks for your interest in our package.

Please be aware that mPdockQ and pDockQ use different scoring formula. You are right about pDockQ as it ranges between 0 and 0.99. however, mPdockQ ranges between 0 and 1.048 because mPdockQ uses different values for L and b from the L and b in pdockQ's formula in the calculation. You can check Equation vii in mPdockQ's paper: https://www.biorxiv.org/content/10.1101/2022.03.12.484089v1.full.pdf Since you said you're modelling trimeric sturctures, mPdockQ will be calculated instead of pDockQ and mPdockQ scores are possible to be greater than 1. Hope it helps.

Yours
Dingquan

from alphapulldown.

Hello Dingquan,

Thank you.
Let me explain the situation a bit more. I have 505 protein targets. Most of these will not/should not interact with my protein dimer of interest. However, out of these 505 protein targets, 503 have mpDockQ score >1. Is that possible? Something seems off to me.

Thanks,
Nirad

from alphapulldown.

Hi,

I see. I suppose the protein dimer models of yours are interacting and with good model qualities e.g. plDDT scores of the residues on the interface of this dimer. If so, then mPdockQ scores will be interfered by this dimer interface. i.e. if the interface between the dimer subunits is bigger and modelled better then the mPDockQ score will remain high no matter the third protein interacts or not.

from alphapulldown.

I see. Yes, the dimer is well known, and the structure has already been solved.
Then should I filter based on iPTM only (>0.6)? I obviously cannot use mpDockQ> 0.23.

from alphapulldown.

I see. Yes, the dimer is well known, and the structure has already been solved. Then should I filter based on iPTM only (>0.6)? I obviously cannot use mpDockQ> 0.23.

Sorry I'm afraid I cannot provide you with suggestions on filtering here as it really depends on the case. iPTM is definitely an important assessment but the other bio-chemical properties reported in the csv table from analysis pipeline could help with your selection. Ultimately, I'm afraid it relies more on the prior knowledge on the biology of these proteins in this case.

from alphapulldown.