Which dataset would most advance AI-driven regulatory genomics?
Open November 16 to December 9, 2026 · Up to 500 words · Open to anyone
Identifying and Building the Datasets for AI-Driven Science is a SeedAI project testing a new way for public and private funders to identify dataset investments that will accelerate AI-driven science. To better understand the needs of the field and to inform the working group, we want to hear from the people working in regulatory genomics who generate and use data.
In 500 words or fewer, describe the following:
Application: What would your field like to be able to do that it can’t do today? For example, predict how a non-coding variant changes gene expression in a specific cell type.
Capability: What would a model need to do well to make that possible, and what measure would define success?
Evaluation: How would you test the capability? Explain what evaluation would be a good stand-in for the capability in the real world.
Data: What training data would help models reach the goal, and roughly how much would be needed?
Questions to consider while preparing your submission
Does it focus on one capability, rather than a list of needs for the field?
Can someone else run the evaluation?
Could a model pass the test just by resembling its training data?
Why would more data help, and how would we know if it didn’t?
What would be the rough cost of a pilot? A full budget isn’t needed.
Scope and submission details
If your idea sits at the edge of regulatory genomics, we still want to hear from you. In your submission, you can provide justification for why you believe your resource question is important for the working group to consider.
Anyone can submit, and we especially encourage submissions from staff scientists, core facilities, industry labs, and institutions beyond the largest research universities. More than one submission is welcome.
Strong submissions may be developed into candidate dataset specifications by the working group. We may summarize submissions without names in published materials. We’ll ask your permission before quoting you or naming you as a contributor.
Submitting does not make you a candidate for funding. Any solicitation that results from this process will be open to everyone.
Questions? Contact Trisha Tucholski at trisha@seedai.org.