Closed prateekbansal97 closed 2 months ago
Hey, thanks for the question.
I think you can do this by parsing a batch of backbones by calling
X_designed, C, S_designed = chroma.design_network.sample(X.cuda(), C.cuda(), sampling_method="potts")
where X has shpe (Nbatch x Nres x 4 x 3) and C has shape (Nbatch x Nres)
Hi @wujiewang ,
Thanks for the reply.
In this case, I generated X and C using X, C, S = protein.to_XCS()
but got an X.shape = [1, Nres, 4, 3]
. How do I make it into size of the batch?
It is simpler than you think, you just need to copy X and C several time and concatenate. something like :
torch.cat( [X] * 10, dim=0 )
The solution you gave is working. Thank you so much!
Hello,
Thanks for Chroma! It is a pretty great tool.
I do have a query - is there a way to design multiple sequences (say 10) for the same input structure? I am new to Chroma and have tried running the design code multiple times with the same structure, but it gives out the same output.
Any help in this matter would be appreciated. Thank you!
Here is the code I am using (same as the one in the README)