Alchemd Benchmark Datasets Repository

Overview

This repository contains input files and benchmark results for JACS molecular datasets employed in Alchemd benchmarking studies. The repository provides standardized molecular structures, protein targets, and transformation pairs derived from established pharmaceutical industry benchmarks for computational method evaluation and performance assessment.

Dataset Provenance

The JACS molecular datasets contained in this repository represent modified structures derived from the prepared_structures available in the OpenFreeEnergy IndustryBenchmarks2024 repository (https://github.com/OpenFreeEnergy/IndustryBenchmarks2024). The input structures have been subjected to additional preprocessing and standardization procedures to ensure compatibility with Alchemd benchmarking protocols while preserving the scientific integrity of the original molecular configurations.

Repository Structure

inputs/
├── BACE/
├── CDK2/
├── JNK1/
├── MCL1/
├── P38/
├── PTP1B/
├── TYK2/
└── Thrombin/
jacs_lig.csv
jacs_pair.csv

Each protein system directory contains three standardized files:

• ligands.sdf - Structure-data format file containing ligand molecules
• protein.pdb - Protein Data Bank file with protein structure
• pairs.lst - Text file containing ligand transformation pairs in format "ligA-ligB"

File Formats

Ligand Files (ligands.sdf)

Structure-Data Format files containing three-dimensional molecular structures of ligands with associated chemical properties and metadata.

Protein Files (protein.pdb)

Protein Data Bank format files containing atomic coordinates and structural information for protein targets.

Pair Files (pairs.lst)

Plain text files listing ligand transformation pairs, with each line containing two ligand identifiers separated by a hyphen (e.g., "ligand1-ligand2"). These pairs define molecular transformations for benchmarking comparative analysis.

Summary File (jacs_lig.csv And jacs_pair.csv)

Excel spreadsheet containing consolidated information and metadata across all molecular datasets in the repository.