There isn't a good stackoverflow/stackexchange-type resource for RosettaRemodel, so this is just a list of bugs we've encountered with some solutions that may or may not work for you. We may eventually port this knowledge over to RosettaCommons or Stack Overflow/Exchange, but for now if you encounter something new and figure out how to solve it, please add it here.
BuildInstructions incompatible! mutable and fixed regions intersect. Fix: Trying to insert residue next to one with a “.” SS assignment; change the adjacent “.” to “L” or “H” or whatever you want it to be.
The sequence position requested was 0. Pose numbering starts at 1. Occurred when I added -repeat_structure flag, with a given structure and adding some denovo build too. Fix: Make sure that symmetry interfaces have at least two residues flexible to build. Try adding de novo residues so it doesn’t build at the end (end on a 95 V L, for example).
The pose does not have residue with chain= , PDBnum=[int]. Occurred when designing sequence in repeat mode with manual resfile syntax. Fix: If specifying design parameters (PIKAA, ALLAAxc, etc) in repeat mode, Remodel will build its own resfile for RosettaDesign on the fly. However it automatically populates any repeat segments as no chain or chain “_”, which disagrees with the native PDB chain (usually “A”) (this is something that can/should be fixed). The stopgap fix is to delete the chain ID from the PDB (select the whole column and replace with a space “ “), and then specify "-chain ” in the flags so that the whole protein becomes chain “”.
Bad loop definition. Fix: when remodeling using the blueprint, cannot have a single residue with flexible backbone - must define at least two consecutive residues with flexible backbones.
ERROR: unable to find desired variant residue: VRT VRT CUTPOINT_LOWER. Encountered when trying to design sequence with manually defined resfile, atompair constraints w/ filter, constrain to start coords w/ weight.
Assert `start > 0` failed. When inserting a pdb structure into another one with Remodel, cannot allow the first 1-2 residues to be flexible. This may be caused only by repeat mode. Fix: Set first two residues to “.” in the blueprint.
failure to read decoy S_00001_0000308_0 from silent-file fold.all.out. All.out files were corrupted and contained W_000* files, which had a different number of energy values in the header and hence the 'I have 31 energy names but I have 25 energy values.’ message. Fix: delete all lines containing W_000 from the file.
Loop closure e.g. “* 0 / 1 closed / attempts”. This can be due to many things. If you are trying to build many loops, that is if you have defined the third column (ELHD/ABEGO) for more than one contiguous segment, you will fail to get a solution if any one of the loops fails to close - your problem becomes harder. Fix: You should examine your loops to see if you are creating hard loop closure problems; you can cut down on the number of noncontiguous loops in a structure you are trying to model at once (combine, or model them one at a time if you can); you can set the max linear chainbreak flag (default=0.07) to allow less strict solutions, e.g.
ERROR: Scorefunction not set up for nonideal/Cartesian scoring. Fix: set the flag
-score:set_weights pro_close 0.