schrodinger.protein.process_nucleotide_db module

This script processes a nucleotide residue database to prepare modified nucleosides for use in mmshare mutation workflows. This includes:

• assigning unique 3-letter residue names to each residue

• checking for valence violations

• adding 5’-monophosphate groups to nucleosides that lack them

• standardizing structure geometry and optionally minimizing

• assigning “similar to” values by fingerprint comparison to standard RNA nucleosides

Output includes a processed residue database file, CSV logs of successes and failures, and a timestamped log file.

schrodinger.protein.process_nucleotide_db.main(options: Namespace) → None

Run the nucleotide database processing pipeline.

Parameters:

options – parsed CLI arguments from parse_args

schrodinger.protein.process_nucleotide_db.parse_args(args=None) → Namespace

Parse command-line arguments for the nucleotide database processor.

Returns:

parsed arguments namespace

schrodinger.protein.process_nucleotide_db.process_nucleotide_db(residues: Iterable[Structure], skip_minimization: bool = False) → PreprocessingOutcomes

Process a nucleotide structure database.

Each residue is assigned a unique auto-generated 3-letter PDB code (ZAA-style).

Parameters:

• residues – iterable of input nucleoside/nucleotide structures

• skip_minimization – if True, skip force-field minimization

Returns:

PreprocessingOutcomes with per-residue results

schrodinger.protein.process_nucleotide_db.get_unique_nucleotide_names(reserved: set[str] | None = None) → Iterable[str]

Generate unique 3-letter codes for custom nucleotides, skipping any names in reserved (built-in nucleotide codes and CSV-assigned codes).

Delegates to get_unique_aa_names so the generated codes follow the same ZAA-style pattern used for custom amino acids.

Parameters:

reserved – set of codes to avoid; defaults to built-in nucleotide names

Returns:

iterator of unique 3-letter code strings

class schrodinger.protein.process_nucleotide_db.CheckValenceViolations(**kwargs)

Bases: PreprocessorStage

Fail any structure that cannot be sanitized by RDKit due to valence errors.

NAME = 'Check valence violations'

class schrodinger.protein.process_nucleotide_db.AddPhosphate(**kwargs)

Bases: PreprocessorStage

Add a 5’-monophosphate group to a nucleoside that lacks one.

Uses an RDKit SMARTS reaction targeting the primary 5’-OH ([OH1][CH2]). Secondary hydroxyls (2’-OH, 3’-OH) are not affected because they are on CH groups, not CH2 groups.

Structures that already contain a phosphorus atom are passed through unchanged. 3D coordinate generation is handled by the subsequent Generate3DStructure stage.

NAME = "Add 5'-phosphate"

class schrodinger.protein.process_nucleotide_db.Generate3DStructure(**kwargs)

Bases: PreprocessorStage

Generate 3D coordinates for structures that lack them.

Structures read from SMILES files or produced by the AddPhosphate RDKit reaction have no 3D coordinates. This stage generates a conformation using the Schrodinger fast-3D pipeline. Structures that already have 3D coordinates (e.g. from a .mae or .sdf input) are passed through unchanged.

NAME = 'Generate 3D structure'

class schrodinger.protein.process_nucleotide_db.StandardizeNucleotideStructure(skip_minimization: bool = False)

Bases: PreprocessorStage

Assign PDB backbone atom names, add hydrogens, check for standard nucleotide name collision, and optionally minimize.

Delegates to NucleotideDatabase.fix_and_validate_res_structure, which internally: checks atom count, assigns backbone PDB names via the OCCC[O]P(=O)(O)(O) SMARTS, rejects built-in standard nucleotide names, adds hydrogens, and runs force-field minimization unless skipped.

NAME = 'Standardize nucleotide structure'

__init__(skip_minimization: bool = False)

class schrodinger.protein.process_nucleotide_db.ValidateMutability

Bases: PreprocessorStage

Verify that the processed nucleotide can be applied as a mutation to a real RNA residue.

Loads an Adenine residue from the built-in rna.bld library as the mutation target and calls infrastructure.mutate_nucleotide. This exercises the same code path used by the mutation workflow and catches structures whose backbone atom-naming is incompatible.

Must run after StandardizeNucleotideStructure, which assigns the correct PDB backbone atom names required by the C++ mutator.

NAME = 'Validate mutability'

__init__()

schrodinger.protein.process_nucleotide_db.apply_nucleotide_analog_properties(processed_residues: list[Structure]) → None

Assign ALIGNS_WITH_PROP and NATURAL_ANALOG_SIMILARITY_PROP to each processed nucleotide.

Parameters:

processed_residues – list of processed nucleotide structures

schrodinger.protein.process_nucleotide_db.get_nucleotide_analog_map(residue_sts: list[Structure]) → dict[str, tuple[str, float]]

Get a map of custom nucleotide names to their closest standard nucleoside, using Canvas fingerprint similarity.

Standard nucleosides are loaded from both rna.bld (A, C, G, U) and dna.bld (T only; A/C/G are shared with RNA and not duplicated). Structures are relabeled with their 1-letter codes so the map values are ready to use as ALIGNS_WITH_PROP values.

Parameters:

residue_sts – processed nucleotide structures; each must have a unique title

Returns:

dict mapping structure title to (1-letter code, similarity score)