Issue with stereochemistry

[karlyy123]

Hello,

I am performing docking experiments with various sugars (glucose, galactose and mannose) using the web server. I see inversion of the stereochemistry, which I do not see with other docking tools (Chai-1, diffdock). As such, I was wondering if there is a way to set up protenix to avoid this inversion. I do not have a computational background and may just be unaware of how to properly configure protenix.

Thanks!

[zhangyuxuann]

@karlyy123 can you give some detailed information or provide the results to us. cc @cloverzizi

[cloverzizi]

Hi @karlyy123 , even when the input 'ref pos' contains specific chiral indications, the output might still display chirality inversion, as noted in the AF3 paper. We're thinking about adding information to the output confidence JSON indicating whether an inversion has occurred. This would help users select the correct sample without inversion. Alternatively, we might explore other solutions to address this issue. Currently, our model cannot prevent inversion.

[ray-berkeley]

Hi @cloverzizi and @zhangyuxuann, I'm also interested in approaches for enforcing ligand/non-protein stereochemistry. Has there (or do you forsee there being) any further work on this issue?