Applying LIS to RFDiffusion Binder Design Filtering

[amelie-iska]

Hi! Thanks for the awesome work. I just thought it might be a good idea to update the pAE method for filtering protein binders designed by RFDiffusion to use the LIS score. For more info, see this issue in the RFDiffusion GitHub:

RosettaCommons/RFdiffusion#211

I'm happy to help out a little and discuss.

[Huilin-Li]

Hi! I am thinking applying LIS to filter my deigned binders. After dl_binder_design/mpnn_fr/dl_interface_design.py, I get a set of .pdb files. However, I find I also need .json files in AFM-LIS.

Could I ask how to calculate LIS and LIA only with .pdb files?

[flyark]

LIS and LIA calculation needs PAE information, which is in json file (colabfold) or pkl file (original alphafold-multimer). Unfortunately, without json of pkl files, LIS and LIA can't be calculated.

…

On Jun 4, 2024, at 9:16 PM, Huilin-Li ***@***.***> wrote: Hi! I am thinking applying LIS to filter my deigned binders. After dl_binder_design/mpnn_fr/dl_interface_design.py, I get a set of .pdb files. However, I find I also need .json files in AFM-LIS. Could I ask how to calculate LIS and LIA only with .pdb files? — Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGHFEW7TXF7EZLF4QHWHKG3ZFZRFJAVCNFSM6AAAAABIEEN5WOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBYGY4DANBXGE>. You are receiving this because you are subscribed to this thread.