Can Boltz-2 be used to predict protein-nucleic acid interactions?

[Tang-pro]

@latkins @jsilter @timodonnell @cunnie
Hi,
Thank you for developing this software. I know Boltz can be used to predict protein-ligand interactions; can it also be used for protein-nucleic acid predictions?

[Galvanized-Heart]

Im pretty sure it's limited to protein small molecule interactions. If you could somehow pass your DNA structure as a SMILES, you maybe could try it. But the context window of the model might limit that. It might also be difficult to get any reliable answers from the interactions like this since it may never have seen DNA binding data like this. If you do try it, definitely have potentials on and check the confidence metric from the model to see how confident it is about the results. Good luck :)

[Tang-pro]

@Galvanized-Heart

Thanks for the tip! I'm going to try that approach.