TADClam: overlapped and hierarchical topologically associating domains detection with community affiliation model

Identifying hierarchical and overlapped topologically associating domains (TADs) from high-throughput chromosome conformation capture (Hi-C) contact maps is an ongoing challenge. Here, we present TADClam, a novel algorithm designed to untangle complex TAD architectures. TADClam employs a weighted community affiliation model to efficiently identify candidate TADs from Hi-C contact maps, followed by an entropy-based filtering procedure to select significant domains. TADClam was evaluated on simulated contact maps with various TAD structures, including disjoint, overlapped, nested TADs, and gaps. Compared to six state-of-the-art TAD callers, TADClam achieved the highest overlapping and weighted similarity ratios, indicating superior performance in detecting complex TAD organizations. When applied to real Hi-C data from human cell lines, overlapped TADs identified by TADClam showed significantly greater enrichment of structural proteins across boundaries. These results demonstrate that TADClam accurately captures complex TAD structures, providing unprecedented insights into 3D genome organization.