FAIR and verifiable: The role of FAIR Digital Objects in trusted AI

Erik Schultes ORCID: http://orcid.org/0000-0001-8888-635X Email: [email protected] Date: October 8, 2025 DOI: https://doi.org/10.5281/zenodo.17257168 https://www.open-science-conference.eu FAIR and verifiable: The role of FAIR Digital Objects in trusted AI FAIR Find me all known low molecular weight inhibitors of the Human p65 Protein. Separate the list based on those that were found in curated databases, from those that were found in self-deposited data archives. Keep track of the license and citation information for each one. If data is relevant, but not public, please provide the contact information for the person I need so I can request the data. Hypothetical search application on FAIR data. Mark Wilkinson, CBGP-UPM/INIA, Madrid 2017 In a world that was FAIR… …you would get a list of inhibitors! …you would get a list of inhibitors! Not a list papers Not a list datasets { "inhibitor": { "name": "Example Inhibitor", "description": "A small molecule that inhibits the activity of the NF-κB p65 subunit.", "identifiers": { "CHEBI": "CHEBI:XXXXX", "PubChem": "12345678", "InChIKey": "ABCDABCDABCDAB-UHFFFAOYSA-N", "SMILES": "C1=CC=CC=C1C(=O)NC2=CC=CC=C2", "molecular_weight": 321.36 }, "target": { "name": "NF-κB p65 subunit", "uniprot_id": "Q04206", "gene_symbol": "RELA", "mechanism_of_action": "Blocks DNA binding of p65 in the nucleus" }, "organism": { "name": "Homo sapiens", "taxonomy_id": "9606" }, "sources": { "curated_database": { "name": "ChEMBL", "record_url": "https://www.ebi.ac.uk/chembl/compound/inspect/CHEMBLXXXXX", "access": "open" }, "self_deposited_data": { "repository": "Zenodo", "record_url": "https://zenodo.org/record/1234567", "access": "restricted", "access_conditions": "Available upon request for non-commercial use" } }, "license": { "type": "CC BY-NC 4.0", "license_url": "https://creativecommons.org/licenses/by-nc/4.0/" }, "citation": { "title": "Identification of inhibitors targeting NF-κB p65 in human cancer cells", "doi": "10.1234/example.doi", "authors": ["Doe, J.", "Smith, A."], "journal": "Journal of Molecular Biology", "year": 2022 }, "restricted_access_contact": { "name": "Dr. Jane Doe", "email": "[email protected]", "affiliation": "Department of Molecular Biology, Example University" } } } Hypothetical JSON Erik Schultes, LACDR 2025 FAIR Find me all known low molecular weight inhibitors of the Human p65 Protein. Separate the list based on those that were found in curated databases, from those that were found in self-deposited data archives. Keep track of the license and citation information for each one. If data is relevant, but not public, please provide the contact information for the person I need so I can request the data. Hypothetical search application on FAIR data. Mark Wilkinson, CBGP-UPM/INIA, Madrid 2017 Finding known low molecular weight inhibitors of the Human p65 protein (also known as RelA, part of the NF-κB complex) can be sourced from curated databases and self-deposited data archives. ### Curated Databases: 1. **ChEMBL**: A well-established database of bioactive molecules, ChEMBL includes several small molecule inhibitors targeting p65/RelA. These molecules are indexed based on their interaction data, activity, and relevance in drug discovery. You can access and download compound data under an open license from the ChEMBL site (EMBL-EBI). 2. **PubChem**: This public resource houses chemical compounds, including those with inhibitory action on p65. Detailed molecular properties, including structure and bioactivity, are available. Licensing for data is open for academic use. 3. **SMMDB**: The SMMDB (Small Molecule Modulator Database) contains detailed information on small molecules modulating targets like p65 in neurological and other diseases. It is a manually curated resource, providing PubMed links and compound details, including molecular weight, 3D structures, and inhibition data【15†source】 【18†source】. ### Self-deposited Data Archives: 1. **ZINC Database**: This includes user-submitted and curated molecules, many of which target NF-κB, including p65. Researchers can deposit their findings here, making it a mixture of community-contributed data. Usage of the database is open, but some entries may require contacting the depositor for further licensing details. 2. **IPAD-DB**: Although primarily focused on Alzheimer’s disease, the IPAD-DB contains data on molecules that modulate related inflammatory pathways, including NF-κB. Licensing details vary, but access is generally open 【17†source】. ### Citation and License Information: - Most curated databases like **ChEMBL** and **PubChem** offer open access with clear academic use licenses. **SMMDB** is similarly freely accessible for non-commercial research, while data in **ZINC** may have different licensing based on individual submissions. - For direct inquiries or requesting non-public data, each database typically provides the depositor's contact, especially in **ZINC** and **IPAD-DB**, where self-deposited entries may have restricted access. For the most comprehensive set of inhibitors targeting p65, I recommend starting with **ChEMBL** and **PubChem** for licensed, peer-reviewed data, and exploring **ZINC** or **IPAD-DB** for community-deposited resources. OpenAI’s ChatGPT-4, October 2024 FAIR Find me all known low molecular weight inhibitors of the Human p65 Protein. Separate the list based on those that were found in curated databases, from those that were found in self-deposited data archives. Keep track of the license and citation information for each one. If data is relevant, but not public, please provide the contact information for the person I need so I can request the data. Hypothetical search application on FAIR data. Mark Wilkinson, CBGP-UPM/INIA, Madrid 2017 OpenAI’s GPT-5 OpenAI’s GPT-5, September 2025 OpenAI’s GPT-5 OpenAI’s GPT-5, September 2025 The FAIR Guiding Principles enhance the ability of machines to do F, A, I and R. What does FAIR mean for AI? https://www.nature.com/articles/sdata201618 https://www.senscience.ai/ Open Data Article https://doi.org/ 10.3389/ fbinf.2025.16341 11 Interactive data explorer https:// www.doi.org/ 10.71728/hw56vj34 Pod cast https:// open.substack.co m/pub/ senscience/p/ staying-ahead-ofsars-cov-2inside? r=58b36l&utm_c ampaign=post&u tm_medium=we b&showWelcome OnShare=false AI driven FAIRification pipelines AI-Ready Data is for AI inputs FAIR-Ready AI will create AI outputs Proposal: FAIR data as the standard for both AI input and output. Trusted AI will consume and produce FAIR Data Yoda iRODs FAIR Data Point Repository Research Object Crate ReBorn Article DeSci Publish CORDRA Nanopub Server Network FAIR^2 Data Portal Emerging technologies supporting FAIR Data Orchestration Yoda iRODs FAIR Data Point Repository Research Object Crate ReBorn Article DeSci Publish CORDRA Nanopub Server Network FAIR^2 Data Portal Emerging technologies supporting FAIR Data Orchestration Proposal: Nanopub-based FDOs as standard I/O for Trusted AI Leiden Laboratory for Nanopublication LLN Leiden Laboratory for Nanopublication LLN AI-Ready Data is for AI inputs https://www.universiteitleiden.nl/en/science/drug-research/metabolomics-and-analytics-centre Metabolomics is the comprehensive, high-throughput analysis of small-molecule metabolites in cells, tissues, biofluids, or organisms under specific physiological or pathological conditions. It aims to identify, quantify, and characterize the dynamic metabolic changes associated with genetic, environmental, or therapeutic factors. ChatGPT5, 30 September 2025 WikiPedia, Creative Commons CC0 1.0 Universal Public Domain Dedication (MAC) Mass Spectrometry What is a nanopublication? What is a nanopublication? Standardized information format What is a nanopublication? • Open standards (HTTP) • Machine-readable (RDF) • FAIR is orchestrated on a global server network (decentralised) • publish • authentication (ORCID) • time/date • Persistent Identifiers •store •search Standardized information format What is a nanopublication? Low levels of 25-hydroxyvitamin D are linked to hypertension and diabetes. Standardized information format • Open standards (HTTP) • Machine-readable (RDF) • FAIR is orchestrated on a global server network (decentralised) • publish • authentication (ORCID) • time/date • Persistent Identifiers •store •search Metadata PID By researcher MAC Nanopubs as FAIR Metadata Tags Standardized information format • Open standards (HTTP) • Machine-readable (RDF) • FAIR is orchestrated on a global server network (decentralised) • publish • authentication (ORCID) • time/date • Persistent Identifiers •store •search 1. has type 2. has location 3. has type-specific metadata https://zenodo.org/records/7782262 Metadata PID By researcher MAC Nanopubs as FAIR Metadata Tags Standardized information format • Open standards (HTTP) • Machine-readable (RDF) • FAIR is orchestrated on a global server network (decentralised) • publish • authentication (ORCID) • time/date • Persistent Identifiers •store •search 1. has type 2. has location 3. has type-specific metadata https://zenodo.org/records/7782262 Single, atomic, semantic unit Nanopubs as FAIR Metadata Tags Standardized information format Metadata for Digitial Objects of type dataset • has type - dataset • has location - PID • has author - researcher • has contributor(s) • has contact point • uses language - English • has domain - metabolomics, hypertension • has publisher - MAC, LACDR, LU • has publication date - xx/xx/xx • has version - x.x.x. • has funder - ZonMw • has project - Hypermarker Nanopubs as FAIR Metadata Tags Standardized information format • has type - method • has location - PID • has author - researcher • has contributor(s) • has contact point • uses language - English • has domain - metabolomics, hypertension • has publisher - MAC, LACDR, LU • has publication date - xx/xx/xx • has version - x.x.x. • has funder - ZonMw • has project - Hypermarker Metadata for Digitial Objects of type method Nanopubs as FAIR Metadata Tags Standardized information format • has type - article • has location - DOI • has author - researcher • has contributor(s) • has contact point • uses language - English • has journal - Nature • has domain - metabolomics, hypertension • has publisher - MAC, LACDR, LU • has publication date - xx/xx/xx • has version - x.x.x. • has funder - ZonMw • has project - Hypermarker Metadata for Digitial Objects of type article Nanopubs as FAIR Metadata Tags Standardized information format • has type - article • has location - DOI • has author - researcher • has contributor(s) • has contact point • uses language - English • has journal - Nature • has domain - metabolomics, hypertension • has publisher - MAC, LACDR, LU • has publication date - xx/xx/xx • has version - x.x.x. • has funder - ZonMw • has project - Hypermarker Metadata for Digitial Objects of type article MAC FAIR Governance Group • Maintain MAC metadata • Maintain MAC vocabulary • Maintain MAC FIP How to make a Nanopub https://nanodash.knowledgepixels.com/project?21&id=https://w3id.org/kpxl/custom/project/terms/MAC-FDO • Semantic modeling • Creating nanopub templates • Minting new FAIR vocabularies • Curating information • “Filling in the form” • Publishing to the server network How to make a Nanopub How to find Nanopub https://nanodash.knowledgepixels.com/project?21&id=https://w3id.org/kpxl/custom/project/terms/MAC-FDO Nanopubs as FAIR Metadata Tags Research Programme Call for Proposals Projects Dataset Software DMP Proposal Ethical Clearance Consortium Agreement Article PI and contributors Budget Method FAIR Funder Framework Margreet Bloemers GO FAIR Foundation Fellow ZonMw. https://osf.io/nxtcj/ Wednesday Nanopub Club Kristina Hettne Aslessa Gambardella Kim Elsink Matthijs Sloep Max van den Boom Elham Memarian Erik Schultes When MAC is FAIR FAIR Find me all Articles published by Amy Harms. When MAC is FAIR FAIR Find me all Methods developed under the Hypermarker project. • Including methods developed by other consortium members When MAC is FAIR FAIR Find me all Instrument Logs related to datasets published in 2025 realted to StayAhead. When MAC is FAIR FAIR Find me all DMPs for projects related to hypertension. • Including DMPs developed anywhere When MAC is FAIR, MAC is AI-Ready Screenshot of MATSU lab Discourse Graph, https://discoursegraphs.com/ LeiBits https://nanodash.knowledgepixels.com/project?13&id=https://w3id.org/kpxl/custom/project/terms/LEIbits https://www.library.universiteitleiden.nl/events/2025/05/connect--interlink LeiBits FAIR Find me all Datasets published at Leiden Univiversity in 2025. Leiden Laboratory for Nanopublication LLN FAIR-Ready AI will create AI outputs Most of the web is AI generated - and thats a big problem. A framework for metadata disclosing the role of AI in content creation. Service providers: •Authoring at scale •Registration at scale •Search at scale Human-! Generated AI-! Assisteded AI-! Generated https://faia.liccium.com/ https://www.sidnfonds.nl/ FDOs as AI output: FAIR AI Attribution (FAIA) ISCC Standard - ISO 24138 MAC Storage Repositories https://osf.io/r625a/files/osfstorage https://osf.io/q9xcu/files/osfstorage Dataset Call for Proposals Research Programme Article Dataset Method Project DMP Method Method Nanopubs for Health-Holland StayAhead Health~Holland’s public–private partnership subsidy https://w3id.org/np/RAof6kN_keiB675cM6GwHmqqlX47ekF-yZ5jKowZsvzSE Research Programme PPP_Allowance_Call_for_Proposals https://w3id.org/np/RArJ3YwHg1LtlbDkHmz3I1BTrJeSuohfsJy4pfmCFnN7w Call for Proposals STAYAHEAD https://w3id.org/np/RAFPOe4OIl_urs-U_7-FostjzcUo8ra6LasfSvGplOvto Project Data Management Plan for STAYAHEAD https://w3id.org/np/RAU1Mo98h2lw3zWADMKx8V7Q03uQNyu2mkTJkRjRQQUOE DMP Methods for project STAYAHEAD https://w3id.org/np/RAjVt8HI8gn9WJwSVIeRRfyVLixF0yKx258RRDM1S2sKI Methods AphaFold2 Dataset STAYAHEAD https://w3id.org/np/RAv6JJe-lvoadhNwml8pzSLs5aOtbU1xbHT5qgFtgwQ-s Dataset ESM Dataset STAYAHEAD https://w3id.org/np/RA9YxiABPRNy3rrJm20p1oanJchBcI22IPpSu36zj2MLs Dataset STAYAHEAD research article https://w3id.org/np/RAnQSVXSxfY3fIdW4f25K_j4imftJ_ijyigLIvS907R8c Article Methods Measurements for Global Polar analysis in project Hypermarker https://w3id.org/np/RAhpepqj9IWw8uiQMIsr9hlpU-zGSUJGpC7xGeZGulZPc