Can contig and unindex coexist in the rfd3 design?

[ruichaomao]

Can contig and unindex coexist? For example, fixing a known motif (with fixed index and structure) + several unindexed residues (only structure fixed) to generate a new scaffold.

[Alexander-the-Evil]

I back this question, have exactly the situation where it is logical to use part of the existing structure, build another part, while including an unindexed motif in this second part. Eager to hear suggestions/solutions!

Additionally, how to specify, in which chain can unindexed residues appear/not appear?

[Ubiquinone-dot]

Yes they absolutely can!
The chain to which unindexed residues belongs is technically leaked by the unindex string;
if you do unindex: 'A1,A2' then the model gets leaked that they both belong to chain A, whereas for unindex: 'A1,\0,A2'; A1 goes to chain A and A2 goes to chain B.

[Alexander-the-Evil]

Hi @Ubiquinone-dot , thank you for the hint!
Indeed, unindex: "/0,B5" worked for me.

@ruichaomao I succesfully combined config and unindex like that:

{ 
    "test": { 
        "input": "./start.pdb", 
        "contig": "A1-154,25-35,A187-279,/0,20", 
        "length": "292-302", 
        "unindex": "/0,B5", 
        "select_fixed_atoms": { 
            "B5": "TIP" 
        } 
    } 
} 

[Cookiemaster33]

I have a comparable problem. Maybe my prompt is a bit more complicated, but I think I handled everything according to @Alexander-the-Evil suggestion.

{
"output_rfd3_11": {
"input": "input.pdb",
"contig": "A1-214,/0,B1-219,/0,D1-60,D288-350,/0,100-150",
"unindex": "/0,/0,/0,E2037",
"select_fixed_atoms": {
"A1-214" : "ALL",
"B1-219" : "ALL",
"D1-60" : "ALL",
"D288-350" : "ALL",
"E2037": "ALL"

		}
    }

}

In runs through, but in the output, E2037 was included into chain A, and not the chain I try to build. I have tried with several amounts of "/0" but no matter what I do, it always end up in chain A.

Any suggestions?

Best

Toby

[Ubiquinone-dot]

It shouldn't be biased toward chain A specifically (saying chain E or whatever as you're doing should work if it's featurized correctly), but on the off chance something weird's going on with the pipelining I'd change the chain you want it to go to to be chain A, i.e you could change the contig you're using to be "100-150,/0,A1-214,/0,B1-219,/0,D1-60,D288-350" and then just "unindex": "E2037".

[Cookiemaster33]

Thanks so much @Ubiquinone-dot.

I have done some trouble-shooting.

The short version is that it seems to work with 2 chains in the pdb + designing a new chain, but not when doing 3 (or more) chains in the pdb + designing a new chain.

Aka this works (E2037 correctly getting integrated into new chain C):

"output_rfd3_11": {
    "input": "input.pdb",
    "contig": "100-150,/0,A1-214,/0,B1-219",
    "unindex": "E2037,/0",
    "select_fixed_atoms": {
        "A1-214" : "ALL",
        "B1-219" : "ALL",
        "E2037": "ALL"

This does not (E2037 incorrectly gets added to chain B):

"output_rfd3_11": {
    "input": "input.pdb",
    "contig": "100-150,/0,A1-214,/0,B1-219,/0,D1-60,D288-350",
    "unindex": "E2037,/0",
    "select_fixed_atoms": {
        "A1-214" : "ALL",
        "B1-219" : "ALL",
        "D1-60" : "ALL",
        "D288-350" : "ALL",
        "E2037": "ALL"

I guess the immediate fix is to put multiple chains together, but might be worth it to look into in the future(?)

Thanks for the advice.

Best

Toby

[ruichaomao]

In my latest test, it worked. Thank you so much! @Ubiquinone-dot @Alexander-the-Evil

[Cookiemaster33]

It now works for me now, too.

In case anyone has comparable problems, these are two changes I made that fixed it. Not sure whether the solution was one of those, or the combination of them.

a) I combined chains in the pdb, so that I only had a total of three chains in the pdb file. 2x target proteins, 1x for unindexed amino acids.

b) For the unindexed amino acids, I changed the residue numbers so that they are labeled continuously from residues 1-10. Previously they were 2037,2046,2101.... etc. Not sure if the problem were the gaps, or that the residue numbers were too high.

Anyways. All set on my end. Thanks everyone again in this thread for the problem solving.