More and more modelers are attempting to use Rosetta to model structures other than those limited to peptide chains, and this will only be increasing. Rosetta has shown great success with RNA modeling, and the modeling of non-canonical amino acids, peptidomimetics, and polysaccharide structures are in active development.
Unfortunately, loading such structures from a PDB file into Rosetta can be challenging. The PDB format is not consistent for non-AA residues, the error-checking of such structures is poor in the PDB, and researchers often ignore all but the most basic standard PDB record types, which in some cases are necessary for proper loading of a file.
Below are some methods of getting your PDB structure to load into Rosetta properly.
The 3-letter code is an unfortunate limitation of the current PDB format. Fortunately, there are a few methods of making sure that your residue is given a 3-letter code in the PDB that will allow it to be loaded into Rosetta.
To load any residue from a PDB file into Rosetta, that residue must have its own unique topology (`.params`) file defined, which needs to be included in the ResidueTypeSet.
This topology file will include the default Rosetta 3-letter code, which should—but will not always—be the same as the PDB standard 3-letter code. Some residues will not already be assigned a PDB 3-letter code, so the Rosetta 3-letter code will necessarily not match. Some of these Rosetta codes will conflict with other 3-letter codes. Thus, it is necessary to have methods for specifying the exact residue type required.
Some classes of residue types can be "turned on" by the use of specific flag options. (See examples below.) For other, more general cases, you can learn about how to "turn on" residue types here:
How to turn on residue types that are off by default
Within a special .codes file, one can specify a list of alternative 3-letter codes and the corresponding Rosetta 3-letter codes, as specified in the appropriate topology files. These .codes files have a very simple format: the first column is an alternate 3-letter code; the second column is the Rosetta 3-letter code; and the (optional) third column is a Rosetta HETNAM record designation for that code.
An example file might contain a line like this:
XXX ALACodes are case-sensitive:
XXX ALA
xxx GLYTo tell Rosetta to consider the alternate codes, simply use the -alternate_3_letter_codes option:
-alternate_3_letter_codes my_codes.codesSeveral example files are present in the database/input_output/3-letter_codes/ directory in the Rosetta database.
One can specify multiple files for inclusion: (Note that if an alternate 3-letter code is present in multiple files or on multiple lines in the same file, the later pairings will overwrite the previous ones.)
-alternate_3_letter_codes my_codes.codes her_codes.codes his_codes.codesIf you do not provide a full path, Rosetta will attempt to read the file from the database/input_output/3-letter_codes/ directory in the Rosetta database. All of the following obtain the same result:
-alternate_3_letter_codes ${ROSETTA}/database/input_output/3-letter_codes/glycam.codes
-alternate_3_letter_codes ${ROSETTA}/database/input_output/3-letter_codes/glycam
-alternate_3_letter_codes glycam.codes
-alternate_3_letter_codes glycamIf the -alternate_3_letter_codes option is given, when Rosetta reads in a PDB it will first check to see if the 3-letter code is found in one of the alternate codes files. If it is, it will use the pairing in the supplied .codes files to translate the alternate code into a Rosetta 3-letter code. If the 3-letter code from the PDB file is not found as a alternate code in the .codes files, Rosetta will will try to use the 3-letter code it found in the PDB file directly.
(For an example of the use of the optional third column in .codes files, see the Carbohydrates section below.)
There are cases when one might have or need to use the same 3-letter code to indicate distinct residue types. In such cases, one can use the PDB HETNAM record type to specify the full name of the base (unpatched) ResidueType needed at that sequence position.
The standard PDB HETNAM record format is deficient for specifying this. Thus, Rosetta uses a "backwards-compatible", modified HETNAM record format.
Standard PDB HETNAM record line:
HETNAM GLC BETA-D-GLUCOSEGLC 3-letter codes in the entire file are beta-D-glucose, which is insufficient, as this could mean twelve different beta-D-glucoses!
Rosetta PDB HETNAM record line:
HETNAM GLC A 1 ->4)-beta-D-GlcpGLC residue at position A1 requires the ->4)-beta-D-Glcp ResidueType.
To load a PDB file with saccharide residues, use the -include_sugars.
Loading of saccharide residues works best using HETNAM records, as described above; however, one can also load (many) PDB files directly from the PDB or those generated from GLYCAM software, (which have their own unique 3-letter-codes,) using the -alternate_3_letter_codes pdb_sugar or -alternate_3_letter_codes glycam flags, as appropriate.
The glycam.codes file makes use of the third column of the list of alternative codes to specify the base ResidueType directly. This is an alternative to the use of HETNAM records and works because a GLYCAM 3-letter code for a saccharide, unlike a PDB 3-letter code for one, is specific for a particular residue type. For example, using the 3-letter code 4GA in a PDB file along with -alternate_3_letter_codes glycam, which includes the following line:
4GA Glc ->4)-alpha-D-GlcpGlc and the record
HETNAM Glc A 1 ->4)-alpha-D-GlcpNote: Rosetta carbohydrate functionality is actively in development; please contact JWLabonte@jhu.edu for assistance/questions.
To load a PDB file with mineral surface residues, use the -include_surfaces flag. This will include all ResidueTypes defined in database/chemical/residue_type_sets/fa_standard/residue_types/mineral_surface/.
To load a PDB file with lipid residues, use the -include_lipids flag.
Note: Rosetta lipid functionality is actively in development and has not been published; please contact JWLabonte@jhu.edu for assistance/questions.
Branching connectivity is defined in PDB files by LINK records. Rosetta will now interpret these LINK records appropriately to build a branching FoldTree by default. Currently, one must add -write_pdb_link_records for them to be written out to any output PDB.