Would Str2Str model work for multichain protein?

[nonsensejoke]

As I read from the paper, this model was trained against single-chain proteins and tested on single-chain proteins. I wonder whether it can work for multi-chain proteins?

Thank you very much.

[lujiarui]

Hi there, thanks for the interest. The method / code may work for multi-chain proteins with some additional modifications to incorporate chain features (such as chain-index and permutation). Also, training on multi-chain should be necessary for the proper performance.