FAIR materials database for NMR Crystallography - A PSDI Webinar

FAIR materials database for NMR crystallography Dr Sathya Sai Seetharaman Computational Scientist, Science & Technology Facilities Council (STFC) Date: 27/11/2025 PSDI Webinar Please use the Q&A and Chat feature if you have a question during this webinar We will answer them at an appropriate time Overview The CCP-NC database NMR Crystallography context Why the database? FAIR gaps NOMAD Oasis (FAIRmat) What have we built? Demos Current developments Work in progress Future plans Next steps Q & A 2 Collaborative Computational Project for NMR Crystallography (CCP-NC) Academia: and more… Industry: 3 What’s in the CCP-NC database? •Plain-text .magres file format •Crystal structure, Calculation metadata, NMR parameters • NMR properties stored ‘per site’ Crystal structures from Density Functional Theory calculations Current database size •~ 5.8k entries •~3.6k organic + ~2.2k inorganic •~2.2k entries linked to unique CSD REFCODES 4 https://www.ccpnc.ac.uk/database https://www.ccpnc.ac.uk/database What’s in the CCP-NC database? •Plain-text .magres file format •Crystal structure, Calculation metadata, NMR parameters • NMR properties stored ‘per site’ Crystal structures from Density Functional Theory calculations Current database size •~ 5.8k entries •~3.6k organic + ~2.2k inorganic •~2.2k entries linked to unique CSD REFCODES 4 https://www.ccpnc.ac.uk/database https://www.ccpnc.ac.uk/database Calculation metadata Crystal structure data Magres – NMR parameters Early use cases for Magres database Avoid re-running/repeating calculations •Quickly check if a magres file already exists for a solid form (e.g. polymorph) •Emphasis on fast negative results—"Not in the database" Guiding experiments •Retrieve typical NMR parameter ranges by material •Helps estimate expected frequency ranges before running an experiment Enable Future Substructure Search •Identify fragments/molecules in the crystal structure •Store the metadata linking back to sites in crystal (link to NMR properties) 5 FAIR Compliance (database version 1) FAIR Area Strengths Current Challenges Findable Accessible •Searchable NMR parameters •Persistent record IDs (e.g., MDBREF 0005210) mimic CSD REFCODEs •Publication DOI, Licenses •No standardised chemical identifiers (free-text searches) •No minted DOIs for records •Substructure search not enabled •Open-access downloads (open science) •Bulk-downloads enabled (suitable for machine learning training sets) •PDDL, CC-BY licenses •No public search API/CLI tools 6 FAIR Compliance (database version 1) FAIR Area Strengths Current Challenges Findable Accessible •Searchable NMR parameters •Persistent record IDs (e.g., MDBREF 0005210) mimic CSD REFCODEs •Publication DOI, Licenses •No standardised chemical identifiers (free-text searches) •No minted DOIs for records •Substructure search not enabled •Open-access downloads (open science) •Bulk-downloads enabled (suitable for machine learning training sets) •PDDL, CC-BY licenses •No public search API/CLI tools 6 FAIR Compliance (database version 1) FAIR Area Strengths Current Challenges Interoperable Reusable •Magres interoperable by design •Magres output support from CASTEP, Quantum Espresso, VASP •Lot more DFT codes can compute NMR parameters •OPTIMADE layer not present •Detailed calculation metadata recorded & displayed •Insufficient metadata output by earlier QE-GIPAW versions •Data provenance could be more detailed 7 CCP-NC Oasis – Findability & Accessibility 12 CCP-NC Oasis – Findability & Accessibility 13 Work in progress – Interoperability & Reusability 14 •Detailed provenance – simulation source files •Workflow linking (CASTEP, Quantum Espresso, VASP) Reusability Interoperability - OPTIMADE •Optimade API endpoint on NOMAD •Custom namespace for CCP-NC NMR quantities •Optimade-maker (NOMAD dependency) Geometry optimisation DFT code outputs Magres file output Evolving use cases – Smarter schema Finding typical shift/EFG parameters •Intramolecular – parameters for given functionality e.g. 15N chemical shifts in amide groups •Intermolecular - correlations with hydrogen bonding or short contacts e.g. CH pointing towards aromatic ring Structurally specific queries Non-molecular framework queries •Identifying structures containing specific linkages e.g. X-O-X tetrahedral connections - Si-O-P where O is bridging two tetrahedra •Purely distance-based - a specific atom (e.g., X) with certain other atoms (e.g., Y and Z) within a defined distance 15 Future Work 16 •Interoperability package to convert from DFT code outputs to magres (temporary delay) •Updating magres parser to detect more DFT output files and linking their workflows DFT parser updates Advancing Findability for Substructure Searching •Workshop with Cambridge Crystallographic Data Centre (CCDC), OPTIMADE, PSDI, and CCP-NC experts •Defined roadmap for developing a user-friendly advanced NMR search •Workflow for extracting detailed substructure information – Soprano, OpenBabel, D_ATA from CCP5 (Thanks to Dr. Chin Yong) Thank You 17 email: sathya-[email protected] Discord: sathya_ss_37382 cosec.stfc.ac.uk