Research / Projects
Mechanistic reasoning
Regulome-R
A reasoning engine that orchestrates genomic models with biological knowledge graphs and literature to explain noncoding variants, moving from “this variant matters” to “this variant disrupts this regulator in this cell type.”
Goal
Variant effect models can say that a variant matters, but not why. Regulome-R aims to connect model predictions with what is already known about regulators, pathways and cell types, and to state clearly when the evidence is thin.
Approach
- Predict. Start from variant effect predictions such as DeepVRegulome.
- Ground. Link each prediction to biological knowledge graphs and the literature.
- Explain. Produce a mechanistic hypothesis a scientist can test, with its supporting evidence.
Status
Regulome-R is in development. Interested in working on it? Join the lab.
Team
Pratik Dutta (lead)