GA4GH Connect session follow up - Agentic Science
0 views
Skip to first unread message
Beatrice Amos
Oct 13, 2025, 4:10:34 PM10/13/25
to Marc Fiume, Work Stream - Data Use and Research Identity (DURI), Clinical and Phenotypic Data Capture, Regulatory and Ethics Work Stream, GA4GH Implementation Forum, Discovery Technical Work Stream, Cloud Work Stream, Data Security Work Stream, MacPherson, Cameron, erdal...@icloud.com, Man Zawati, Sasha Siegel, Munoz Torres, Monica, Davidcs, boris.g...@smrtr.com.au, Alexanderkanitz, Work Stream - Genomic Knowledge Standards (GKS)
Dear Colleagues,
Thank you all for contributing to the GA4GH Connect session: Agentic Science: Enabling AI Agents to Reason with Biomedical Data via GA4GH Standards - what a productive and engaging discussion! A big thank you to Marc Fiume and Cameron MacPherson, and our guest speakers, Sasha Siegel, Mónica Muñoz Torres, David Steinberg, Boris Guennewig and Ma'n H. Zawati for an excellent session, and to all participants for their valuable contributions and engaging discussions. Your input made this a truly meaningful conversation.
Thank you all for contributing to the GA4GH Connect session: Agentic Science: Enabling AI Agents to Reason with Biomedical Data via GA4GH Standards - what a productive and engaging discussion! A big thank you to Marc Fiume and Cameron MacPherson, and our guest speakers, Sasha Siegel, Mónica Muñoz Torres, David Steinberg, Boris Guennewig and Ma'n H. Zawati for an excellent session, and to all participants for their valuable contributions and engaging discussions. Your input made this a truly meaningful conversation.
‼️ Please take a moment to review the meeting notes and share any edits or additions to help keep the summary accurate and complete. If you notice any mistakes, have resources to add, or want to share more details, please feel free to jump in and update the notes - we welcome your input! A link to the recording is available in the linked meeting notes.
________
Key Takeaways
🔶 AI as a Foundational Disruptor: AI is transforming genomics from descriptive to predictive science, requiring urgent GA4GH standards for data readiness, ethical use, and governance to handle its rapid pace and risks.
🔶 Collaboration and Inclusivity Essential: Success demands industry-academia partnerships, cross-workstream alignment, and emphasis on ethics/equity to democratise access and ensure responsible AI deployment.
🔶 Balance Innovation with Safeguards: While embracing agentic science and hallucinations for discovery, prioritise QC via synthetic data/RAG, human oversight, and metadata provenance to maintain trust and accuracy.
Key Takeaways
🔶 AI as a Foundational Disruptor: AI is transforming genomics from descriptive to predictive science, requiring urgent GA4GH standards for data readiness, ethical use, and governance to handle its rapid pace and risks.
🔶 Collaboration and Inclusivity Essential: Success demands industry-academia partnerships, cross-workstream alignment, and emphasis on ethics/equity to democratise access and ensure responsible AI deployment.
🔶 Balance Innovation with Safeguards: While embracing agentic science and hallucinations for discovery, prioritise QC via synthetic data/RAG, human oversight, and metadata provenance to maintain trust and accuracy.
________
Meeting Summary
The session centered on establishing a new foundational AI workstream to address the rapid integration of AI in genomics and health data, emphasising its disruptive potential for accelerating research, data analysis, and discovery while ensuring responsible, ethical, and equitable use.
Presenters presented lightning talks on practical AI applications, including variant annotation using GPT models with retrieval-augmented generation (RAG) to eliminate hallucinations, the Narrative Project for privacy-preserving data descriptions, agentic scientists leveraging GA4GH standards for hypothesis generation and workflow execution, and BRCA variant analysis.
A collaborative FigJam brainstorm explored AI's impacts over the next 3 years, such as agentic science, improved accessibility, and risks like data silos and biases, alongside GA4GH's role in developing standards, governance, and cross-workstream partnerships.
Discussions highlighted challenges like data quality control (QC), hallucinations, synthetic data generation, metadata standards for provenance, human oversight, and the need for industry-academia collaboration to make genomics data AI-ready.
________
________
Action Items
🔶 Organise an open strategy session to condense FigJam themes, prioritise products/study groups, and align with other workstreams for AI readiness.
🔶 Establish the GA4GH AI Work Stream and get it off the ground (establish charter, leadership, principles, and road map) - Recruit and nominate co-leads for the AI workstream; interested individuals should contact Sasha directly.
🔶 Conduct a landscaping session with other GA4GH workstreams to define territories, enable AI integration, and address painful interoperability issues like shared data models.
🔶 Explore partnerships with external groups (e.g. ML Commons) for metadata standards and cross-pollination on consent/provenance.
🔶 Organise an open strategy session to condense FigJam themes, prioritise products/study groups, and align with other workstreams for AI readiness.
🔶 Establish the GA4GH AI Work Stream and get it off the ground (establish charter, leadership, principles, and road map) - Recruit and nominate co-leads for the AI workstream; interested individuals should contact Sasha directly.
🔶 Conduct a landscaping session with other GA4GH workstreams to define territories, enable AI integration, and address painful interoperability issues like shared data models.
🔶 Explore partnerships with external groups (e.g. ML Commons) for metadata standards and cross-pollination on consent/provenance.
________
Please take a moment to complete our Exit Survey: https://forms.ga4gh.org/t/qjHm9R1EGmus
Your feedback is invaluable and helps us improve future GA4GH Plenary and Connect meetings, ensuring they continue to provide the best possible experience for our community.
Thanks again for your engagement and contributions!
Best,
Beatrice
Beatrice Amos
Work Stream Manager
Reply all
Reply to author
Forward
0 new messages