Chromosomes exhibit remarkable organizational complexity. Vast strands of DNA fold into compact, highly ordered structures that balance accessibility to active genomic regions with the sequestration of less frequently used sequences. Deciphering this three-dimensional organization has long been a challenge, as chromosomes are massive, dynamic systems that require both experimental and theoretical approaches to fully understand.
In a recent PNAS publication, researchers José Onuchic and Vinícius Contessoto from Rice University introduced FI-Chrom, a powerful new method for reconstructing three-dimensional chromosome structures from experimental data. FI-Chrom leverages Hi-C chromosome contact maps, which divide chromosomes into segments, or “beads,” each representing roughly 50,000 DNA base pairs. These maps capture how frequently different beads come into proximity, providing probabilistic interaction data rather than direct spatial coordinates.
Transforming these probability-based relationships into accurate 3D structures presents a major computational challenge, given the hundreds of thousands of beads and millions of interactions involved. FI-Chrom overcomes this by using inverse statistical mechanics to generate spatial arrangements that satisfy all Hi-C constraints simultaneously.
“We previously had chromosome maps that contained implicit three-dimensional information, but we were interpreting them in two dimensions,” said José Onuchic, corresponding author of the study. “FI-Chrom allows us to convert these data into realistic 3D chromosome models.”
Developed by first author Antonio Oliveira Jr., FI-Chrom applies a maximum entropy framework and is trained solely on experimental Hi-C data, without imposing assumptions about chromosome structure. Remarkably, the resulting models naturally reproduced known features of chromatin organization, such as compartmentalization and knot minimization.
Beyond static structure, FI-Chrom also enables researchers to infer chromosomal dynamics. By analyzing interaction frequencies across cell populations, the team demonstrated that chromatin loops form transiently rather than existing as fixed architectural features.
