QUESTION: Does anyone know of a software similar to PepComposer for computational peptide design?

Question

[alexille5640]

PepComposer (https://www.ncbi.nlm.nih.gov/pubmed/27131789) computationally generates peptides binding to a given protein surface. It only requires the target protein structure and an approximate definition of the binding site. However, the website appears to be down right now.

Does anyone know of a software which is similar?

[ravi.thakkar369]

Hey did you find any similar program or server ? I found one server PinaCoalda, but it needs an Ligand peptide and receptor peptide complex.

Answer 1

[matteoferla]

Rosetta is great for protein design experiments of various finds. It does not have this specific functionality, but does allow you to engineer interfaces or dock peptides to a surface and so forth. It has its own scripting language and a few of its methods use 6-mer and 9-mer libraries of fragmented PDB structures to do protein design (c.f. Rosetta Remodel). At its core it does "hacked force-field" calculations, so it is doing stuff kind of properly so can allow side-chains to move (e.g. those spinning lysines, arginines and glutamates) etc. which for surface design experiments is really good.

However, the server options are understandibly limited: Rosie.

If you know where exactly you want your peptide you could simply use remodel: add a new first chain with only two N-terminal residues placed strategically in a few locations and write a blueprint to extent this chain. The "two" is because the first residue needs to be fixed, but the next one along needs to be wiggled. This is what is a blueprint:

    1 A .
    2 A L PIKAA A
    0 X L ALLAA
    0 X L ALLAA
    0 X L ALLAA
    0 X L ALLAA

Five long will be relatively quick (10 minutes single core), and you can build from the best scoring solutions (most negative ddG). --generic_aa G will need to be set. In the above blueprint I asked for any of the all amino acids, but I could have been selective. Also the L means loop. D for helix might be better if you want a rigid peptide (in which case the picked amino acids can be more logical, given that only 2/3.5 will bind).

In summary, the bad news is that I am not suggesting an alternative, the good news is that I am suggesting something more complicated but that may give better results.