gemmi

The gemmi drg module generates monomer restraint dictionaries for crystallographic refinement. It combines chemical rules with statistical knowledge derived from AceDRG tables.

Core Workflow:

1. Read Monomer Definition: Typically a CCD-like CIF file.
2. Build Molecular Graph: Analyzes atoms, bonds, and valence context.
3. Consult AceDRG Tables: Looks up bond/angle targets and sigmas.
4. Emit Restraints: Produces categories suitable for refinement workflows (bonds, angles, torsions, chirality, and planarity).

Note that the produced values are empirical/statistical refinement targets (ideal values and sigmas/esds), not QM geometry optimizations.

Gemmi is a library and set of programs designed for structural biology and macromolecular crystallography (MX). It provides tools for handling various crystallographic data formats and biological models.

Core capabilities include:

• Macromolecular Models: Working with PDB, PDBx/mmCIF, and mmJSON files.
• Refinement & Small Molecules: Handling CIF files for refinement restraints and small molecule models.
• Reflection Data: Processing MTZ and mmCIF formats.
• Symmetry: Managing crystallographic symmetry and symmetry-aware analysis.
• 3D Grids: Handling electron density maps, masks, and MRC/CCP4 formats with crystallographic symmetry.

Supported Platforms & Languages:

• Platforms: Linux, Windows, macOS, and WebAssembly (for web browsers).
• Languages: C++14 (core), Python 3.8+ (bindings), and partial C and Fortran 2003 interfaces.

Gemmi supports four primary formats for storing crystallographic reflection data (Miller indices and associated values):

• MTZ files: The most common format used in macromolecular crystallography.
• Structure factor mmCIF files: Used for data archiving in the Protein Data Bank (PDB).
• XDS ASCII format: Data generated from processing diffraction images with XDS.
• Small molecule structure factor CIF files: Typically identified by the .hkl file extension.

Gemmi is available through several interfaces depending on your workflow:

• C++ Library: The core implementation (requires C++14).
• Python Bindings: For integration into Python-based scientific workflows.
• Command-line Tools: Standalone utilities for processing files.
• WebAssembly: Partial bindings for web-based applications.
• Fortran: A Fortran 2003+ interface (currently in progress).

Additionally, Gemmi provides online tools such as GemmiMOL (a molecular viewer) and various data visualization projects.

Gemmi provides extensive tools for 3D grid (map) manipulation:

• grid.hpp: Core 3D grid structures used by CCP4 maps, cell-method search, and HKL data.
• blob.hpp: Finds maxima or "blobs" in a Grid (map), similar to CCP4 PEAKMAX.
• floodfill.hpp: Implements the flood fill (scanline fill) algorithm for a Grid, assuming periodic boundary conditions and 6-way connectivity.
• isosurface.hpp: Extracts isosurfaces from 3D grids using the marching-cubes algorithm.
• mc_tables.hpp: Contains lookup tables for the marching-cubes algorithm.
• ccp4.hpp: Support for the CCP4 map and mask format.

Gemmi aims to implement a route for generating a single, deterministic, idealized conformer for well-restrained small molecules from restraints. This process is more complex than hydrogen placement because it involves globally coupled constraints like torsions, chirality, ring closure, and nonbonded contacts.

The proposed pragmatic workflow is:

1. Build an internal-coordinate tree.
2. Seed rigid fragments (small rings, aromatic systems, planar groups, peptide/nucleic templates).
3. Place the remaining graph by walking bonds using bond/angle/torsion restraints.
4. Enforce chirality during placement.
5. Run restrained Cartesian minimization to clean up closure errors and clashes.
6. Use a fallback embedding/minimization step for difficult rings/macrocycles.

Gemmi follows the IUCr dictionary for chirality-related terminology:

• Sohncke groups (non-enantiogenic): The 65 space groups containing only operations of the first kind (rotations, rototranslations, translations). Chiral structures crystallize in these groups.
• Enantiomorphic space groups (chiral groups): The 22 groups that form 11 enantiomorphic pairs. Note that chiral structures can also crystallize in 43 of the achiral (non-enantiomorphic) space groups.

Symmetry operations can be combined using the multiplication operator *. When operations are combined, they are treated as symmetry operations and the result is automatically wrapped to the $[0, 1)$ range.

Operations

• Combination: op1 * op2 combines two operations.
• Inversion: op.inverse() returns the inverse.
• Wrapping: op.wrap() reduces the translation modulo 1.

Note that op.wrap() is not automatically applied to individual operations upon creation (e.g., x,y+1,z is not automatically reduced to identity), but it is applied during combination to ensure results remain within the standard crystallographic range.

import gemmi

op = gemmi.Op('-y,x-y,z+1/3')

# Combine operations
combined = gemmi.Op('x-y,x,z+1/6') * '-x,-y,z+1/2'

# Invert and wrap
inverted = combined.inverse()
wrapped = inverted.wrap()

The PDBx/mmCIF specification includes "alternate" versions of data types, ranges, and enumerations (e.g., _pdbx_item_type vs _item_type). These alternates are typically interpreted as stricter deposition constraints.

To apply these stricter criteria during validation, use the use_deposition_checks option.

• In Python: Set use_deposition_checks as a keyword argument in the cif.Ddl constructor.
• In C++: Set the use_deposition_checks member variable directly.
• In CLI: Use the --depo flag with the gemmi validate command.

# Python example
validator = cif.Ddl(use_deposition_checks=True)