Rosetta module selection

[munir-yousef]

Hello everyone, I am a new Rosetta user and was wondering which rosetta module is best suited for protein - peptide structure prediction using de novo amino acid sequence inputs?

[gezmi]

Hi,

You can use PatchMAN for that - the newest version, exclusively using PyRosetta can be found here: https://github.com/Furman-Lab/PatchMAN/tree/python
(we will merge this branch to the main soon)

There are 2 scripts, one for running the pipeline using SLURM, the other is running it "locally". Especially for larger proteins (and receptor backbone minimization), you will need quite a few CPU-s to do the latter.

Please let us know if you need any help.

Also tagging @orafurman

[lyskov]

@munir-yousef also we are working on a Patchman ROSIE app which should become available ~soon. I can ping you when it is ready if you are interested, please let us know.

[munir-yousef]

Hello,

Thank you for your prompt response! I have a few more questions that I will share below. I apologize if this is not the right forum, please let me know if there is somewhere else I should ask these questions or if there is any documentation you can point me to where i can find the answers.

• Do all of the products (Patchman and those in the ROSIE app) fall under the PyRosetta license?
• My understanding is that RosettaTTAFold All-Atom can be used to predict the structure of protein:peptide complexes, are the above modules better suited for these tasks?
• Can you please share which modules also predict affinity?
• Which modules, if any, predict for non natural amino acids or canonical only?
• Which modules can be used to predict binding affinity after swapping out an amino acid in the sequence?
• Any modules or software that can model cyclic or non natural amino acid peptides?

Thank you in advance for your time and guidance!

[gezmi]

Hi,

PatchMAN is not a Rosetta module per se, but a pipeline that incorporates PyRosetta, so it does fall under the PyRosetta license. Other apps in ROSIE usually use the C++ Rosetta
There is currently no "Rosetta module" global docking - there are pipelines that can use Rosetta for some steps.

Our advice is to first use AF3, RFAA, Chai1, etc softwares. In many cases they do already work. If not, the slower global focking protocols, e.g. the state-of-the-art PatchMAN can be a good alternative.

Rosetta can handle quite a few NCAA-s and post-translational modifications, therefore PatchMAN also can (and the underlying FlexPepDock protocol, used for local refinement in the pipeline). However, we know that Rosetta's energy function is not great for NCAA-s. Also, if an NCAA is missing, you need to add parameters for that.

Specifically for peptide-protein complexes, you can look at FlexPepBind to predict affinity changes. Also, running FlexDDG, or cartesian relax (look at Rosetta documentation for these) can be useful.

I am not 100% but I think that you can build cyclic peptides but you cannot dock them to a protein with Rosetta (but I can be mistaken).