Is it possible to predict the structure by considering the effect that the two proteins have on each other?

[chersiakingdom]

When there are two proteins whose structure you want to predict, is it possible to predict the structure by considering the effect that the two proteins have on each other?

For example, does the method of folding two monomers described in the README file take into account the effect they have on each other? (COMMAND : --fasta_paths=monomer1.fasta,monomer2.fasta --> monomer1 and monomer2 are folded in consideration of their effect)

Or.... is that command just for user convenience? then, Is there another way to achieve what I want?

---

As another curiosity, I wonder if it is possible to put a pdb file (tertiary structure) in the input file rather than the primary structure amino acid sequence. (I think it probably won't,)

Thanks for your help. It will be of great help to me.

[andycowie]

The multimer model does model the interaction between the two proteins - it attempts to predict how they will form a complex. In this case you should follow the instructions for folding a heteromer, and specify both peptide sequences in a single fasta file - https://github.com/deepmind/alphafold/blob/main/README.md#folding-a-heteromer.

It is not currently possible to supply a pdb instead of a fasta sequence for one of the proteins.

[chersiakingdom]

@andycowie

Hello. Thank you for your answer. but your words confuse me a little.

We know that multimer means two or more peptide chains joined together, and folding into a heteromer takes into account the interaction of the two chains.

However, what I am curious about is whether the multimer model takes into account the effects of two different 'proteins' on each other. (Not for two different peptide chains, but for the folding of different proteins)

E.g, https://github.com/deepmind/alphafold/blob/main/README.md#folding-a-heteromer. When using the above method to fold the heterolog A2B3, can you see A as a single protein and not as a peptide chain? What I mean is that, when I want to fold the both protein of T1050.fasta and the protein of T1051.fasta in consideration of their interaction, can I write AB as AB in one fasta file? (where A is the sequence of T1050, B is the sequence of T1051)

When you say that there is an A1B1 protein and an A1B2C3 protein (two separate proteins), If you want to fold individual proteins in consideration of their influence on each other,

A B A B B C C C

Do you mean to create and fold one fasta file like the one above?

I'm trying to search on my own, but it's not clear. could you please answer me?

[SiriusFuenmayor]

I am also interested in this question, for example for determining binding between two proteins, any information will be greatly appreciated.

[andycowie]

Hi @chersiakingdom, you are correct, AlphaFold Multimer predicts complexes from individual chains. If you wish to look at interaction between complexes, the fasta file format you outline is the correct way to input this.