How Do We Crack the Genome’s Hidden Structure? A Numerical Journey of Trial and Error
- Day - Time: 16 December 2026, h.12:00
- Place: Area della Ricerca CNR di Pisa - Room: C-40
Speakers
Referent
Abstract
We often think of DNA as a linear sequence of letters. But inside our cells, this incredibly long molecule is folded into a complex three-dimensional structure, where a piece of DNA is located can matter almostas much as what it says. Among the building blocks of this architecture are Topologically Associating Domains (TADs), genomic neighborhoods that bring certain regions of DNA into close contact while keeping others apart. Detecting these structures, however, is not simply a matter of “looking” at the genome: it is a challenging computational problem that requires turning complex 3D genomic measurements into the right numerical representation.
Identifying these structures requires strong modeling assumptions, since agnostic approaches, including standard AI methods, can fail to capture the underlying structure of the data.
In this seminar, through a story of trials and (several) errors, we will explore modeling assumptions, highly parallelizable numerical methods, and graph theory, on a journey through genome architecture guided by one central question: “What is the right way to formulate the problem?”