abmptools 1.15.1

Pre/post-processing and analysis toolkit for ABINIT-MP Fragment Molecular Orbital calculations

ABMPTools (ABINIT-MP Tools)

A Python toolkit for pre-processing, post-processing, and analysis of Fragment Molecular Orbital (FMO) calculations with ABINIT-MP.

Features

IFIE/PIEDA Analysis (getifiepieda, anlfmo, cpf2ifielist, getcharge)

• Distance-filtered IFIE tables for target fragments or molecules
• Fragment–fragment interaction matrices (1:1, 1:N, N:1, N:N)
• Time-series IFIE from MD-FMO trajectory snapshots
• SVD-based interaction decomposition
• Charge extraction from ABINIT-MP logs

CPF Management (cpfmanager, convertcpf, generate_difie, log2cpf)

• Parse and write CPF files (versions 4.201, 7.0 MIZUHO, 10, 23)
• Version conversion between CPF formats
• Residue-based CPF extraction
• Dynamic IFIE (DIFIE) averaging across MD snapshots with mean/σ statistics
• Generate CPF from ABINIT-MP log files

FMO Input Generation (generateajf, pdb2fmo, udf2fmo, setfmo, addsolvfrag)

• Auto-generate AJF input files from PDB structures
• Fragment assignment for proteins and molecular assemblies
• Solvation fragment addition
• Support for sp2 fragmentation and various basis sets

File Format Conversion

• CIF → PDB/XYZ (readcif) with symmetry operations
• ABINIT-MP log → fragment config (log2config, ajf2config)
• PDB editing and serial AJF generation (pdbmodify, ajfserial)

GROMACS ↔ OCTA COGNAC Conversion

• udf2gro: Convert OCTA UDF files to GROMACS format (.gro, .top, .mdp, .itp)
• gro2udf: Convert GROMACS files to OCTA UDF format (supports --from-top mode)

Geometry Optimization (geomopt)

• MacePdbOptimizer: MACE/ASE-based PDB structure optimization
• OpenFFOpenMMMinimizer: OpenFF force-field minimization via OpenMM
• QMOptimizerPySCF: Quantum chemistry optimization with PySCF

Amorphous Structure Building (amorphous)

• Multi-component amorphous system construction using Packmol and OpenMM

Supported ABINIT-MP Versions

• ABINIT-MP v1: Rev.10–23
• ABINIT-MP v2: Rev.4–8

Installation

Editable install is recommended for day-to-day use and development:

pip install -e .

Non-editable install (e.g. for production deployment):

pip install .

--user is usually unnecessary; pip handles both virtual environments and system Python appropriately.

Installation runs make to compile the optional Fortran shared library for accelerated IFIE/PIEDA reading. If gfortran is not available, the install still succeeds without Fortran acceleration.

Requirements

• Required: Python 3.8+, numpy, pandas
• Optional: UDFManager (OCTA COGNAC), gfortran, OpenBabel, PySCF, ASE, OpenMM, Packmol

Quick Start

# Extract IFIE for fragment 10, within 8 Å
python -m abmptools.getifiepieda --frag 10 -d 8.0 -i calculation.log

# Generate AJF input from PDB
python -m abmptools.generateajf -i protein.pdb -basis 6-31G* --method MP2

# Convert log to CPF
python -m abmptools.log2cpf -i calculation.log -o output.cpf

# Create DIFIE-averaged CPF from trajectory
python -m abmptools.generate_difie -i traj-xxx.cpf -t 1 10 1 -f 1-100 -np 4

# Convert UDF to GROMACS
python -m abmptools.udf2gro.cli -i system.udf -o output

# Convert GROMACS to UDF
python -m abmptools.gro2udf.cli -i system.gro -t system.top -o output.udf

# Build an amorphous mixture (e.g. 50 ketoprofen molecules, density 0.8 g/cm^3)
python -m abmptools.amorphous --smiles "OC(=O)C(C)c1cccc(C(=O)c2ccccc2)c1" \
    --name ketoprofen --n_mol 50 --density 0.8 --output_dir ./ketoprofen

Use -h with any module for full option details.

Documentation

• User Manual — CLI options, output formats, and workflow examples
• Architecture — Class hierarchy and design overview
• Developer Quickstart — Setup and code conventions
• I/O Spec — File format specifications
• gro2udf / udf2gro — GROMACS ↔ OCTA conversion
• geomopt / amorphous — Optimization and structure building

Testing

pytest tests/ -v                     # 671 tests across 30 files
pytest tests/ -v -k molcalc          # specific module
pytest tests/test_regression.py -v   # regression tests (60 bundled + 16 gated)

See tests/TEST_COVERAGE.md for details.

Regression Tests

tests/test_regression.py compares current CLI output against reference fixtures stored in tests/regression/reference/ (generated from the pre-refactor state). This guards against behavior drift during refactoring.

Covered tools: generateajf, log2cpf, convertcpf, udf2gro, gro2udf, and getifiepieda.

Developer-only tests: the 16 getifiepieda regression cases require external sample data (the internal abmptools-sample repository) at:

../abmptools-sample/sample/getifiepieda/
├── 6lu7-multi-fmolog/    (extracted from abmptools-fmolog-sample.tar.bz2)
├── cd7-fmolog/
├── 6m0j-pb-fmolog/
└── xyzfile/

These tests are automatically skipped when the data is not available, so public CI runs are unaffected.

Samples

Each sample/ subdirectory contains input data and a run.sh script:

cd sample/generateajf && bash run.sh
cd sample/log2cpf && bash run.sh
cd sample/generate_difie/TrpCage && bash run.sh
cd sample/convertcpf && bash run.sh

Author

Koji Okuwaki