RoseTTAFold All Atom

The rf-all-atom API provides an interface to access the RoseTTAFoldAllAtom folding tool. The source code and additional documentation can be found on the github page https://github.com/baker-laboratory/RoseTTAFold-All-Atom.

This RosettaFold All Atom tool is capable modeling structures of protein protein complexes, protein dna complexes, and protein ligand complexes, as well as proteins with covalent modifications. Protein RNA complexes are currently broken in the base RFAllAtom code. We will add logic for running with RNA once this issue has been resolved.

The inputs for the rf-all-atom are protein sequences, dna sequences, ligand sdf files, and a json string that defines covalent modifications.

The input arguments are as follows:

--fasta-file: The path to the protein fasta file(s)

--dna-fasta: The path to the dna fasta file(s)

--ligand-sdf: The path to the ligand sdf file(s)

--covalent-bonds: A json string defining the covalent bonds. See the RoseTTAFold-All-Atom github page linked above for more details.

--gpu-type: The type of GPU to use. If more than about 500 protein residues or about 50 nucleic acid residues are present a A100 gpu should be used. Otherwise the default t4 should be sufficient.

Outputs:

The output is a tarball with the file labeled “results.pdb” being the predicted structure. The other files mostly represent intermediates and logs which are useful for debugging.