fusong-ju / profold Goto Github PK

Hello,
While predicting the distance map, Did you consider only the Carbon Beta atom while predicting distance map ? What does it mean by CBCB? Or did u consider all backbone atoms like N, Carbon Alpha, C, and Carbon Beta?

It's a great honor to read your work, and I really hope to see the output of “MSA encoder” and "Co-evolution aggregator" (the input of "Distance estimator", not the raw MSA).
I would be grateful if you could tell me how to get it！

As depicted, I have some .fasta files which represents sequences of nucleic acid, like:

2122_1|Chains A[auth V], B[auth W], C[auth X], D[auth Y], E[auth Z]
MSEAAQTLDGWYCLHDFRTIDWSAWKTLPNEEREAAISEFLALVDQWETTESEKQGSHAVYTIVGQKADILFMILRPTLDELHEIETALNKTKLADYLLPAYSYVSVVELSNYLASGSEDPYQIPEVRRRLYPILPKTNYICFYPMDKRRQGNDNWYMLSMEQRRELMRAHGMTGRKYAGKVTQIITGSVGLDDFEWGVTLFSDDALQFKKLVYEMRFDEVSARFGEFGSFFVGTRLPMENVSSFFHV

how can I prepare the input ? or can I directly use the fasta as input? Any idea?

It's a great honor to read your article, but I really hope to see your training network. I would be grateful if you could provide it！

Hi there,

Thanks for sharing your work with us.

Could you please suggest how to generate the aln format MSA that fits the model?

Thanks a lot.

Hi Fusong,

It is exciting to read your new method on modeling co-evolution. ProFOLD here is wrapped for structure prediction, but I think the results of CopulaNet should be also useful to construct fitness landscape. Is that possible to obtain joint probability distribution of amino acid residues in the MSA from the output of ProFOLD? I do find a numpy .ngz file in the output folder, but have no idea whether it is feasible or how to construct the joint probability distribution using this file. Thank you very much.

Best

Zhixing