treeclimbR pinpoints the data-dependent resolution of hierarchical hypotheses

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Hierarchical data structures are prevalent in biology, including phylogenetic trees and cell type classifications.
• Analyzing these structures at multiple resolutions is crucial for understanding complex biological systems.
• Existing methods may lack the flexibility to capture latent signals across different data granularities.

Purpose of the Study:

• To introduce treeclimbR, a novel computational tool for analyzing hierarchical biological entities at various resolutions.
• To develop a data-driven method for identifying significant features within hierarchical structures.
• To provide a flexible framework for exploring biological associations across multi-resolution datasets.

Main Methods:

• treeclimbR employs an approach to identify multiple candidate signals within hierarchical trees.
• The method pinpoints specific branches or leaves containing features of interest.
• Performance is evaluated against existing methods using synthetic and real-world biological data.

Main Results:

• treeclimbR demonstrates superior performance compared to current methods on synthetic datasets.
• The tool effectively analyzes diverse biological datasets, including microbiome, microRNA, single-cell cytometry, and RNA-seq data.
• The approach successfully captures latent signals and identifies features of interest across different resolutions.

Conclusions:

• treeclimbR offers a robust and flexible solution for analyzing multi-resolution hierarchical biological data.
• The tool facilitates a thorough inspection of entities and aids in uncovering novel biological associations.
• This method is valuable for researchers working with complex genomic and biological datasets.