Disulfidizer API

The Disulfidizer API runs the Rosetta “Disulfidizer” protocol. The disfulfidizer protocol attempts to stabilize a protein by making disulfide bonds. This method has been used widely with atomic-resolution success for de novo designed small proteins, and less extensively for larger proteins. The user may restrict what sets of residues the protocol searches for disulfides between. The protocol will return protein structures for as many disulfides and combinations thereof it found to be stabilizing, as well as a scorefile list of structures. If it is not possible to make reasonable disulfide bonds for the given backbone, no models will be returned. See Notes for more details on the disulfidizer protocol.

Examples

Command Line Examples

Run disulfidizer on an input structure allowing any residue to form disulfides

lev engine submit disulfidizer input.pdb

Run disulfidizer on an input structure in permissive mode to allow more permissive disulfide bond formation and relax output models.

lev engine submit disulfidizer input.pdb --permissive --run-relax

Run disulfidizer on an input structure allowing residues 5-8 to form disulfides with residues 30-32

lev engine submit disulfidizer input.pdb --residue-set-1 "5 6 7 8" --residue-set-2 "30 31 32"

Run disulfidizer with max disulfides set to 1

lev engine submit disulfidizer input.pdb --max-disulfides 1

Python Examples

Run disulfidizer on an input structure allowing any residue to form disulfides

from engine.disulfidizer.client import DisulfidizerClient

client = DisulfidizerClient()
job_id = client.submit(pdb_path="input.pdb")

Run disulfidizer on an input structure in permissive mode

from engine.disulfidizer.client import DisulfidizerClient

client = DisulfidizerClient()
job_id = client.submit(pdb_path="input.pdb", permissive=True)

Run disulfidizer on an input structure allowing residues 5-8 to form disulfides with residues 30-32

from engine.disulfidizer.client import DisulfidizerClient

client = DisulfidizerClient()
  job_id = client.submit(
  pdb_path="input.pdb",
  residue_set_1=[5,6,7,8],
  residue_set_2=[30, 31, 32])

Inputs

  • --pdb-file
    • Input PDB file – a cleaned and/or relaxed PDB file
    • Prepare a clean PDB using the Clean PDB API, especially when using residue sets in command.
    • Levitate strongly recommends using the Prepare or Relax tool in Levitate Bench to preprocess models before uploading to this API. This will provide a more stable scoring baseline and will make interpreting results easier.

Options

  • --residue-set-1 (str)
    • Space separated residue numbers which should be allowed to form disulfide bonds with the residues in --residue-set-2
    • See Notes for more details
  • --residue-set-2 (str)
    • Space separated residue numbers which should be allowed to form disulfide bonds with the residues in --residue-set-1
    • See Notes for more details
  • --permissive (boolean)
    • Enable more permissive disulfide bond formation instead of standard disulfide detection
    • See Notes for more details
  • --max-disulfides (int)
    • Set the maximum number of disulfides that can be created per model
    • Recommended setting this value to 1 if you are modeling a larger protein
    • default = 1000
  • --run-relax
    • Run Rosetta Cartesian Relax on all models output by disulfidizer

Outputs

  • models (directory)
    • PDB files stabilized with the addition of disulfide bonds
    • Number of models will vary based on the number of feasible disulfide bond arrangements.
    • Models will be named in the form input_0001_####.pdb, where #### is an index (0001, increasing).
    • If a model named input_no_new_disulfides_found.pdb is present, it indicates that no new disulfides were found.
  • score.sc
    • Textfile containing Rosetta scores for each generated model
    • If no new disulfides were found this file will have a line of text indicating that.
    • See Notes for more information

Notes

Rosetta Disulfidizer Protocol

To summarize the documentation:

  • This protocol scans a protein and builds disulfides that join residues in one set of residues with those in another set. GLY and PRO residues are ignored. Residues too close in primary sequence are ignored. Potential disulfides are first identified by CB-CB distance, then by mutating the pair to CYS, forming a disulfide, and performing energy minimization. If the energy passes a cutoff, it is compared with a set geometric criteria from a database of known disulfides in the PDB. If the “distance” along this metric passes a cutoff, the pairing is considered a valid disulfide bond.

  • Once valid disulfides are found, they are combinatorially added. For example, if disulfides are identified between residues 3 and 16 and also between residues 23 and 50, the following configurations will be found: 1. [3,16] 2. [23,50] 3. [3,16],[23,50]

  • The Rosetta Disulfidizer protocol described in further detail here (paper) and here (documentation).

Using residue sets

  • This is useful for a case like the inside of a beta-sandwich protein: one set is each interior face of the protein.

  • If you choose to supply residue set data, if there are already disulfides in your model, be sure that they are entirely inside or outside of your residue sets. Putting half a pre-existing disulfide in a residue set is forbidden behavior.

  • Uploaded PDBs must start at residue 1 to match this numbering scheme. If you do not intend to use residue sets, PDB numbering requirements are less strict.

  • If you supply no residue set data, the code will just look across the entire protein. This may be problematic if the protein is large (you may get too many results to sort through) or if you specifically want focussed results.

  • Levitate recommends selecting residue sets, for ease of interpretation of the results.

Using permissive mode

  • Levitate has chosen best-practice settings on your behalf. We have bundled several into one user facing option: standard versus permissive disulfide detection.

  • Standard detection uses Rosetta’s default settings and will provide the most plausible disulfides. The downside is that it will find no plausible disulfides for some protein backbones/positions.

  • Permissive detection tunes the disulfide score and geometric criteria cutoffs to more permissive settings.

    • This means that Rosetta is much more likely to find candidate disulfides.
    • Many hits will be lower quality.
      • Many hits will appear as “not disulfides” in external software such as PyMOL – they will be oriented too poorly to be shown as a proper disulfide.
    • Rosetta may find a tremendous number of disulfides (or combinations thereof) for some backbones, making the results harder to interpret.

Levitate recommends standard settings for conservative designs featuring the most likely disulfides, or permissive settings for hypothesis generation and creating large data sets for cross referencing with other tools.

Output File interpretation

Results can be downloaded if and only if a job has succeeded – DONE state, see the list command below. If no viable arrangements are identified, the job will have failed (status FAILED) and attempting to download its results will give an error message stating that there are no models.

The model numbering is irrelevant to quality. The models are ordered in a way such that the first model (input_0001_0001.pdb) will contain the most N-terminal disulfide in the set, and the last-numbered model will contain the mostly C-terminal numbered disulfides.

Model quality can be assessed via scores. Broad documentation on interpreting scores can be found here. The score.sc file is a space-delimited data table, padded for easier reading. It can be parsed with pandas dataframes, excel, or your tool of choice. In the score.sc file, pay particular attention to the total_score and dslf_fa13 columns (the second and third). The total_score is Rosetta’s overall grade for a model. dslf_13 is a measurement of just the quality of the disulfides in the model. This is a total score for all disulfides, so if there are several it adds them together. In both cases, these scores are more-negative-is-better. The total_score should have a negative value for good models, although this will depend enormously on the quality of the input model. The dslf_13 score will have a small negative value in an ideal case but a small positive value is common and may be acceptable. A large positive value is a sign of a bad disulfide and should be treated cautiously. The dslf_13 score for an individual disulfide can be found by looking at the energy table at the bottom of the pdb file. This breaks down the per residue energy for each amino acid in the protein. The dslf_13 score is found in the 14th column. And each CYS is in the pair is awarded half the total score (so double the value you see for 1 residue to get the total score for the pair).

Disulfidizer will output all combinations of disulfides that it can (see the introduction above). As a consequence, the disulfide score for any individual disulfide is present for the model that contains only that disulfide. You can also find this data in the per-residue score tables at the end of each PDB file. These are also in a space-delimited tabular format that can be imported into your analytics tool of choice.

Rosetta will also report the disulfides into the PDB models in two ways:

  • the standard SSBOND record
SSBOND     CYS A    4    CYS A   66                                       3.94
  • an additional comment record
REMARK PDBinfo-LABEL:   66 DISULFIDIZE

Example PDB data:

SSBOND     CYS A    4    CYS A   66                                       3.94  
...ATOM records omitted
REMARK PDBinfo-LABEL:   66 DISULFIDIZE

These can be parsed to get a quick overview of where disulfides went in an entire return dataset.

Treatment of existing disulfides

Disulfidizer is set up to discover disulfides even where they already exist. As a consequence, models with existing disulfides may have them “rediscovered”, which may appear as a repacking or rotamer change.

References

Accurate de novo design of hyperstable constrained peptides

Rosetta Disulfidize

Rosetta Energy Terms

Updated: