The only way I could the loop modeling and docking to work simultaneously is to have the “correct” fragment library files. The only way I found to obtain the correct files was to merge both fasta sequences into one continuous one. Yes this seems odd, but it works for me. The other part of the puzzle is to have both proteins in the .pdb file with continuous (at least unique) numbering.
Results so far have been expected.
Initially, I generated frag libraries for one of the proteins by submitting only its fasta sequence. Libraries were generated, but they were rejected by rosetta (even though I only wanted to flex loops in this one protein).
Ultimately, I was able to flex loops in both proteins while docking using this method.
In summary, my pdb has two proteins numbered 29-342 and 400-505. I have a fasta file containing 328 residues, all in a “single” sequence, of which I submitted to Robetta for library creation.
Let me know if I can add more details. Yes, I’ve spent a good bit of time with trial and error