Tree-Based Co-Clustering Identifies Chromatin Accessibility Patterns Associated With Hematopoietic Lineage Structure.

Area of Science:

• Genomics
• Hematopoiesis
• Computational Biology

Background:

• Chromatin accessibility varies across hematopoietic cell types.
• Existing methods lack the ability to associate accessibility with the hematopoietic differentiation tree structure.

Purpose of the Study:

• To develop a novel co-clustering approach to associate chromatin accessibility variation with the hematopoietic differentiation tree.
• To capture regulatory patterns and improve clustering of cell types based on accessibility.

Main Methods:

• Utilized ATAC-seq data from the ImmGen consortium.
• Developed a tree-aware co-clustering algorithm accounting for differentiation lineage.
• Validated findings using publicly available ChIP-seq datasets.

Main Results:

• Identified 12 cell type clusters and 20 genomic locus clusters capturing ~80% of accessibility variation.
• Demonstrated that cell type clusters reflect coherent components of the differentiation tree.
• Showed that the novel algorithm produces less dispersed cell type clusters compared to hierarchical and k-means methods.

Conclusions:

• The novel tree-based co-clustering method accurately associates chromatin accessibility with hematopoietic differentiation.
• The findings provide insights into gene regulation across hematopoietic lineages.
• A Python package, chromcocluster, is available to implement the presented algorithms.