DSWB - FAIR Federation Webinar - eLwazi by Dr Sumir Panji

FAIR-ification and Federation within H3ABioNet and the eLwazi Open Data Science Platform –Chapter 1 https://elwazi.org/ https://dsi-africa.org/ Data Science Without Borders webinar –2nd October 2025 Sumir Panji (as part of H3ABioNet) Challenge –Lack of African Genomics data https://h3africa.org/ H3ABioNet and summary of H3Africa Data Challenges with Human genomics data H3Africa Data Archive Availability of H3Africa Data www.h3abionet.org #h3abionet Catalogue design H3ABioNet SAB Meeting, 3rd –5th August 2020 Biorepository data dumps Archive data dumps Public access Registered access DBAC access Studies Datasets Participants Biospecimens Ethics approval POPIA compliance Security checks Email triggers Registrations Processing https://catalog.h3africa.org/ FAIR? Findable, Accessible, Interoperable, Re-usable https://data.europa.eu/doi/10.2777/1524 Why is FAIR important? Any suggestions and thoughts from the audience? Why is FAIR important? https://op.europa.eu/s/sniv FAIR-ness assessment of H3Africa Data H3Africa data FAIR Assessment H3Africa data FAIR Assessment https://catalog.h3africa.org/ H3Africa data FAIR Assessment •100% on the Assessment for the level of Essential, 76.19% on Assessment Non-Essential with an overall Assessment score of 87.8% FAIR Indicators for H3Africa Data FAIR Indicators for H3Africa Data Reuse Licence for Human Genomics data – Data Use Ontology What else should be made FAIR? Any suggestions / thoughts from the audience? What else should be made FAIR in H3ABioNet? H3ABioNet Workflows project –FAIR? https://www.h3abionet.org/tools-and-services/workflows H3ABioNet Workflows project –FAIR Conclusions •Limiting factor in genomics is not data generation, increasingly multi-dimensional data - data analysis and interpretation are bottlenecks •To apply tools and techniques such as ML, well labelled data is required •Curation of data / good dataset stewardship is essential towards adding value for data to enable findability and re-usability •The FAIR landscape has changed from policy discussion to implementation with better metrics and tools now available for assessing FAIR-ness of different resources and infrastructures •Better understanding and guidance on implementing FAIR and its metrics for a range of outputs now available •Skills and knowledge on FAIR gained within H3ABioNet over this period –moving from making outputs FAIR to now creating “born-FAIR” outputs •Knowledge, skills and experience gained being applied within other African data health research projects e.g. eLwazi Open Data Science Platform (https://elwazi.org/), as part of the The Data Science for Health Discovery and Innovation in Africa (DS-I Africa) Initiative (https://dsi-africa.org/) Acknowledgements H3ABioNet is funded by the NIH Common Fund Award / NHGRI Grant Number U24HG006941 Questiuons / Quick Activities - 5 -10 mins max Choose 1 of the activities below to give it a quick go: Activity 1: •Navigate to: https://tinyurl.com/dswb2025 •Use the Australian Research Data Commons FAIR Data Self-Assessment Tool to get a quick idea of how FAIR your data set is, or what is needed to make a dataset FAIR Activity 2: •Navigate to: https://tinyurl.com/fairsw •Follow the instructions to clone the repo and run it on a github project you have, what is missing according to the output? FAIR-ification and Federation within H3ABioNet and the eLwazi Open Data Science Platform –Chapter 2 https://elwazi.org/ https://dsi-africa.org/ Data Science Without Borders webinar –2nd October 2025 Sumir Panji (as part of eLwazi ODSP) NIH DS-I Africa Harnessing Data Science for Health Discovery and Innovation in Africa Research Hubs: Advance and demonstrate feasibility of data science research and innovation to improve health in Africa Training: Increase capacity for data science research in Africa ELSI Research: Explore Ethical, Legal, and Social Implications of data science research from an African perspective and contribute to policy discussion on the continent Open Data Science Platform & Coordination Center: Facilitate the development of a trans-African network of data scientists Image and text extracted from: https://commonfund.nih.gov/africadata NIH DS-I Africa https://dsi-africa.org/resources/infographics eLwazi partners SUN: de Oliviera UCT: Mulder Wits: Hazelhurst UMaur: Baichoo UVRI/Makerere: Kayondo, Djingo UKart: Fadlemola USTTB: Doumbia UCSC: Paten Broad: Lawson, Loreth UChicago: Grossman EBI: Burdett 7 African partners (from H3ABioNet), 3 USA, 1 UK eLwazi aims: The need we aim to fulfil: •Projects have multiple partners and datasets •Integrating different data types •Need to find the data, identify tools, run analysis •Need access to storage and compute The solution needs to be feasible in the African context eLwazi aim: To develop an African Open Data Science Platform and associated resources, to support the Harnessing Data Science for Health Discovery and Innovation in Africa (DS-I Africa) consortium and beyond Aims to be a flexible, scalable platform enabling the implementation of data science for health, that is relevant to the African context How we support the DS-I Africa consortium •Storing data, including reference datasets •Serving metadata •Developing data models •Data management, cleaning, QC •Data harmonization Data •Make software stack available •Developing/adapting new tools •Building containerized workflows •Customizing for computing environment Tools •Making components available on multiple environments •Easing access to computing facilities •Ensuring security is in place as required Computing •Curriculum development •Hosting training materials •Training on data science methods •Training on eLwazi platform Training Data portals Metadata catalogue REDCap support Hosting protocols, SOPs, guidelines Helpdesk User support groups Data working groups FAIRPlus Data Set Maturity Model https://fairplus.github.io/Data-Maturity •Content-related: What is reported in the Dataset (data) & the Dataset Descriptor (metadata). •Representation and format: How the data object & metadata object are represented and formatted. •Hosting environment capabilities: What capabilities of the hosting environment that enables and supports the use of FAIR data. eLwazi Data Jamboree 2023 31 Attendees from 9 research hubs / projects joined for the week. Focused on: 1. Introduction of FAIRification resources •FAIRPlus •RDA Framework •Data Use Ontology 2. Brainstorming Metadata Data Catalogue, Initial Design 3. Explore solutions for common harmonisation challenges https://elwazi.org/trainings/24 eLwazi Data Jamboree 2024 39 Attendees from 19 research hubs / projects joined for the week. Focused on: 1. Examined FAIRification Journey •FAIRification Framework •FAIRcookbook 2. Enabling greater visibility of all DSI Africa datasets and fostering of collaborations via the Metadata Catalogue 3. Engaged with real tools in Data Harmonisation https://elwazi.org/trainings/29 FAIR-ification across DS-I Africa Next Jamboree –February 2026 Focus on: 1. Reviewing FAIR progress over the last 3 years 2. Identify and promote end-to-end use cases 3. Advancing metadata catalogue features In the meantime, eLwazi solutions for FAIRification: a. Metadata Catalogue (F, A, R) b. Support & Further Development of Metadata Harmonisation Tool (developed by HE2AT) (I, R) c. Data Transformation Tool (I, R) eLwazi Metadata Catalogue https://catalog.elwazi.org/#/datasets Challenge https://www.datalaw.africa/law/search_compare/ eLwazi ODSP Infrastructure WG Interoperability and Standards Enable international data sharing Promote sharing across the translational continuum Encourage technology-enabled federated approaches Promote interoperability GA4GH aims to... Build standards and APIs to enable responsible sharing of data for health benefits Addressing the fact that datasets may or may not be able to move Standards, tools and APIs facilitate finding data and tools and executing analysis in multiple computing environments Image credits: https://www.ga4gh.org/news_item/ga4gh-standards-in-a-global-learninghealth-system/ GA4GH Data Repository Service •Provides a standardised set of data access methods that are agnostic to cloud infrastructure •Allows for data access regardless of storage location or how the data is managed Image credit: https://www.ga4gh.org/product/data-repository-service-drs/ GA4GH Data Connect •Works with any data that can be serialised as an array of JSON objects •Supports federation •Can be implemented across a large variety of data stores Image credit: https://www.ga4gh.org/product/data-connect/ eLwazi ODSP follow-up GA4GH 2025 hackathon •Inclusive, collaborative discussions and planning by eLwazi research software engineers, partners from the GA4GH community, DNAStack and Gen3 teams 4 streams defined for the eLwazi ODSP GA4GH 2025 follow-up hackathon: •Stream 1 –eLwazi ODSP Catalogue, DNAStack and Publisher / Explorer and Data Connect integration •Stream 2 - Identity gaps for integration with DNAStack and Gen3 and Ilifu on prem (around workbench) •Stream 3 - Case studies •Stream 4 –Implementation of GA4GH standards in TREs eLwazi ODSP follow-up GA4GH 2025 hackathon eLwazi ODSP follow-up GA4GH 2025 hackathon #eLwaziHackathon2025 eLwazi ODSP integration with DNAStack https://elwazi.omics.ai/collections Putting it together Cloud & HPC Environments Azure, GCP, AWS and Local •DRS for data access •WES for compute •TRS for workflow sharing •Workspaces Dockstore Methods Portal & Data Catalog Workspaces Local Compute and Storage Nodes AWS Azure Global context https://healthwise-consortium.org/ •University of Cape Town, PI Nicky Mulder •University of the Witwatersrand, PI Scott Hazelhurst •University of the Western Cape/Ilifu, PIs Mattia Vaccari, Rob Simmonds, Russ Taylor •University of Stellenbosch, PI Tulio de Oliviera •University of Mauritius, PI Shakuntala Baichoo •Uganda Virus Research Institute, PIs Jonathan Kayondo, Daudi Jjingo •University of Khartoum, PI Faisal Fadlemola •USTTB, University of Bamako, PI Seydou Doumbia •Broad Institute, PIs Jonathan Lawson, Christine Loreth •EMBL European Bioinformatics Institute, PI Tony Burdett •University of California, Santa Cruz, PI Benedict Paten •University of Chicago, PI Robert Grossman Acknowledgements NIH DS-I Africa program, Grant U2CEB032224