molecular docking

[wangjiawen2013]

Hi,
AlphaFold3 supports molecular docking now, does Protenix support molecular docking ? I mean performing molecular docking by Protenix itself, without the help of other downstream tools such as autodock, gold, discover studio and so on.

[zhangyuxuann]

@wangjiawen2013 Hi, if you mean cofolding protein and ligand or other type molecule, ProteniX have already support it, you can try ProteniX server. If you mean fixing protein structure and do rigid docking like vina, ProteniX do not support this feature yet! By the way, you said that “AlphaFold3 supports molecular docking now”, do you mean the scenario of Rigid docking? I think the docking success rate metrics reported in AF3 is also in the scenario of cofolding.

[wangjiawen2013]

@zhangyuxuann
Yes, I mean cofolding here.
I find another github repository named protenix-dock https://github.com/bytedance/Protenix-Dock, what does protenix-dock do and what's the relationship between protenix cofolding and protenix-dock ?

[zhangyuxuann]

@wangjiawen2013 ProteniX-dock was basically finalized a year ago, but we only recently had the time to open source it. Like Vina, ProteniX-dock is based on MC search with a scoring function (PScore) with biophysical and empirical terms including AMBER/GAFF2 forcefields, VdW, Hydrogen bonds, pi-stacking, cation-pi, torsion strain, hydrophobicity, ionic interactions and rotatable bonds.

The ProteniX series will have a series of work around biological complexes, including traditional methods and AI based methods. Protenix-dock is one of them.

[wangjiawen2013]

@zhangyuxuann Pleas have a look at this interesting discussion:
ccsb-scripps/AutoDock-Vina#411 (comment)

If I understand correctly, Protenix cofolding is AI-based and Protenix-dock/vina is traditional molecular docking methods, right ?

[zhangyuxuann]

@zhangyuxuann Pleas have a look at this interesting discussion: ccsb-scripps/AutoDock-Vina#411 (comment)

If I understand correctly, Protenix cofolding is AI-based and Protenix-dock/vina is traditional molecular docking methods, right ?

Yes!

[wangjiawen2013]

@zhangyuxuann Sorry to trouble again. I am studying on the interaction between a protein complex with eight subunits and a ligand. After cofolding using Protenix, how to measure the docking quality (or "docking strength") between the ligand and the complex ?
DockQ is a tool for assessing the quality of protein interfaces in model structures against given references, so I think we should not use DockQ here, because there is no reference here, we just want to discover whether the ligand can bind to the complex and quantify the binding interactions (or affinity).