search

generatebio / chroma

A generative model for programmable protein design
Apache License 2.0
627 stars 74 forks source link

Caption data #1

Closed lhallee closed 7 months ago

lhallee commented 7 months ago

Hello,

Amazing work! Curious if the captions used to train the ProCap will be made available. Best, Logan

aismail3-gnr8 commented 7 months ago

Thanks! The captions used to train ProCap are taken from the descriptive text for PDB structures, as well as UniProt functional comments for individual chains. For more details, see Section T.2 of the Supplemental Information of our paper.