AI for scientific discovery · Ongoing collaboration

Long-context genomics

Investigating whether long-context DNA models encode linkage disequilibrium in collaboration with genomics researchers at CNR Italy.

Ongoing collaborationLead · Gianluigi VitaleLast updated · 30 August 2026

Overview

Investigating whether long-context DNA models encode linkage disequilibrium in collaboration with genomics researchers at CNR Italy.

Problem

It is not yet clear which population-genetic structures long-context DNA models represent and whether those representations recover biologically meaningful linkage patterns.

Approach

The project combines long-context machine-learning systems methods with genomics expertise to test for encoded linkage disequilibrium.

Results & current status

  • The collaboration and experimental program are active.
  • The work connects long-context representation analysis with domain-grounded genomics validation.

Technical details

This research brings together sequence modeling, long-context evaluation, representation analysis, and population-genetics expertise.

Reproducibility

Results will be published only when the study has a stable methodology and collaborators have agreed on the public release.