GENOMIC SUB-COMPARTMENT PREDICTION USING ENSEMBLE HI-C

Abstract

Genome organisation, i.e. how such a huge DNA (around 6 feet) is organised inside such a small cell nucleus of around 5 μm, has always been an area of interest because of its biological significance, like disease diagnosis and treatment, influence on gene expression, designing synthetic genomes, studying evolutionary relationships and many more. Challenges like 3d structure complexity, lack of low-resolution genomic data, and variability across cell types within an organism pose difficulty in the perfect study of the same. Experimental methods, e.g. Hi-C and Lamina-DamID, which capture chromatin interactions among DNA fragments and what regions are near nuclear lamina, respectively, are being used to perform such studies, revealing biological facts and patterns associated with the genome. High Coverage Hi-C experiments are costly and available only for GM12878 human cell lines. Usually, the experimental data (Hi-C) is averaged over millions of cells, so we get an ensemble view of the genome. Research has given genomes different names at various resolutions, such as TADs, loops, compartments, etc. Compartments are further classified as active and inactive, which help us understand what genes are active during various processes in the body; they also reveal how genes interact with each other. Researchers leverage these experimental data to derive compartments and subcompartments within and across various cell lines. To understand the cell-to-cell variability of genome organisation, we need to study the organisation at the individual cell level, but sparsity in data is the major challenge there. This work is also aligned with exploring compartments and subcompartments classification across the genome in the GM12878 cell line using various online experimental data.