Introduction
In the dynamic landscape of the genome, chromatin organization plays a pivotal role in regulating gene expression, cell differentiation, and development. Among the various structural features of chromatin, loops formed by interactions between distant genomic regions are crucial for bringing regulatory elements into close proximity with their target genes. However, the mechanisms governing the formation and resolution of these loops, especially in the context of cellular changes over time, remain inadequately understood. This project aims to shed light on the dynamics of loop appearance and disappearance across different stages of cell differentiation and the cell cycle using high-resolution time-series Hi-C data. By modifying the HiCCUPS algorithm for enhanced loop detection in time-series datasets, we seek to unravel the intricate patterns of chromatin reorganization, offering new insights into the molecular underpinnings of cellular function and identity.
Methods
Datasets
Our study utilizes three high-resolution time-series Hi-C datasets, focusing on various cell differentiation processes and the cell cycle. These datasets were selected for their comprehensive coverage of chromatin interactions over time, providing a unique opportunity to observe loop dynamics in different biological contexts. By employing data from the 4DN network, we ensure a high standard of data quality and comparability across different experimental setups.
HiCCUPS Modification and Loop Detection
HiCCUPS, a widely used algorithm for detecting chromatin loops, was adapted to meet the specific requirements of our time-series analysis. The original HiCCUPS framework was enhanced to incorporate temporal data, allowing for the detection of loops that appear or disappear over time. This involved recalibrating the algorithm’s sensitivity to changes in loop strength and incorporating statistical methods to validate these changes across multiple time points.
Loop Appearance and Disappearance Analysis
To systematically analyze loop dynamics, we defined specific criteria for classifying loops as appearing or disappearing. Using Spearman correlation and Bonferroni correction for statistical validation, we developed a metric called ‘trend accuracy’ to assess the reliability of our loop classifications. This approach allowed us to capture the nuanced patterns of loop formation and resolution in a quantifiable manner, providing a foundation for deeper biological interpretation.