FAIRifying the TEMPL Pipeline: Applying FAIR Principles to a Template-Based Protein–Ligand Pose Prediction Baseline

FAIRifying the TEMPL Pipeline Applying FAIR Principles to Scientific Software Jozef Fülöp Martin Šícho Wim Dehaen CZ-OPENSCREEN, Department of Informatics and Chemistry Faculty of Chemical Technology University of Chemistry and Technology, Prague October 20, 2025 What is TEMPL? A Template-Based Protein–Ligand Pose Prediction Baseline Uses maximal common substructure (MCS) alignment Constrained conformer generation for 3D structures Optimized for rapid prediction within known chemical space CPU/GPU adaptive for different computational resources The FAIR Guiding Principles F Findable Data and software are easy to discover for both humans and computers through persistent identifiers and rich metadata. A Accessible Data and metadata can be retrieved using standardized protocols and are available under clear licensing terms. I Interoperable Data uses standard formats and vocabularies to enable integration with other data and systems. R Reusable Data and software are well-described with clear provenance and licensing to support future research and applications. Learn more at GO-FAIR → Findable: Multiple Persistent Identifiers Persistent Identifiers (DOIs) Four separate DOIs ensure permanent, citable references for software, datasets, benchmarks, published paper Rich Metadata README.md, citation File Format (citation.cff) provides machine-readable bibliographic information for automatic discovery Multiple Indexing Indexed across GitHub, Zenodo, JCIM, ChemRxiv, and Software Heritage for maximum discoverability Software DOI 10.5281/zenodo.16890956 Dataset DOI 10.5281/zenodo.15813500 Benchmarks DOI 10.5281/zenodo.16875932 Published Paper (JCIM) DOI 10.1021/acs.jcim.5c01985 Accessible: Open Licenses & Multiple Access Methods Open Licensing MIT License for code - Code is freely available for reuse, modification, and distribution with minimal restrictions. CC BY 4.0 for data & documentation - Data and documentation are openly shared with proper attribution requirements. SPDX (= Software Package Data Exchange)  - headers in all Python files, machine readable clear license for every file (code)  # SPDX-FileCopyrightText: 2025 TEMPL Team # SPDX-License-Identifier: MIT Multiple Access Methods Command-line interface (CLI) for direct use Web application (hosted e-infra cz) Python API for integration & programmatic access Docker containers for deployment (docker file on GitHub) Automated data download from Zenodo (zenodo_get), one command .env setup (script) Interoperable: Standard Formats & Cross-Platform Standard Formats PDB SDF JSON CSV SMILES Community Tools RDKit for molecular chemistry ESM-2 for protein embeddings Standard PDB conventions Cross-Platform Deployment Python 3.12 Docker and Kubernetes support for containerized deployment Hardware detection (CPU/GPU adaptive) uv - fast Python package and project manager SDF Input: Output: (Protein Data Bank)  3D protein structures (Structure-Data File)  MDL Molfile format SMILES (Simplified Molecular Input Line Entry System) with 3D mol coordinates (predicted pose) (structured data/metadata)  for easy parsing (tabular results) SMARTS patterns for molecular substructures Reusable: Documentation & Reproducibility 📜Clear Licensing MIT License for code enables maximum reusability. CC BY 4.0 for data and documentation ensures proper attribution. SPDX headers in every Python file for machine-readable licensing.  📚Comprehensive Documentation README.md, CONTRIBUTING.md, CHANGELOG.md, and module-specific guides.  Executable Jupyter Notebook (workflow reference - prefer executable examples), - - help in CLI  🔍Provenance Tracking Timestamped output directories: 🔄Reproducibility Features Exact dependency pinning (RDKit 2024.09.6), automated environment setup, Docker & Kubernetes (e-infra CZ), GitHub Actions CI/CD, test suite (pytest), implemented automated benchmarking using CLI using: PDBBind (time-split) & Polaris datasets output/templ_run_20250926_145408_2etr/ ├── 2etr_top3_poses.sdf ├── 2etr_all_poses.sdf ├── 2etr_pipeline_results.json └── 2etr_template.sdf Software Heritage: Long-Term Preservation Permanent archival of source code independent of platform changes Ensures TEMPL remains accessible even if GitHub becomes unavailable Provides version tracking and complete development history Enables reproducibility and verification of research results Free service supporting open science and research transparency TEMPL Archive on Software Heritage What's Next: Research Software Directory About RSD & bio.tools (Elixir EU) The Research Software Directory and bio.tools are platforms that catalogs research software and make it discoverable to the global research community. Increases visibility among researchers Establishes TEMPL as a recognized research tool Improves citation metrics and impact Connects with other research software projects Next Steps - List TEMPL in Research Software Directory - Register with bio.tools  Research Software Directory https://research-software-directory.org/ bio.tools (Elixir EU) https://bio.tools/