Target inference from collections of genomic intervals

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Identifying recurrent genomic regions in noisy biological data is a significant challenge.
• Existing methods often struggle with data variability and identifying regions of varying lengths.

Purpose of the Study:

• To introduce Cores of Recurrent Events (CORE), a novel computational approach for detecting significant recurrent genomic regions.
• To demonstrate the utility of CORE in analyzing tumor heterogeneity and identifying cancer-related genomic aberrations.

Main Methods:

• CORE formalizes the explanation of observed genomic data using "core" intervals.
• It employs a combinatorial optimization procedure, selecting the number of cores (depth) based on statistical significance.
• The method is designed to accommodate cores of widely varying lengths.

Main Results:

• CORE successfully identifies significant recurrent genomic regions in both synthetic and real-world biological data.
• Application to single-cell tumor DNA copy number profiles revealed tumor population phylogeny and subpopulation-defining features.
• Analysis of comparative genomic hybridization data from breast cancer samples identified recurrent copy number aberration regions.

Conclusions:

• CORE provides an effective computational solution for finding recurrent genomic events in noisy datasets.
• The method has significant applications in understanding tumor evolution and defining cancer-specific genomic alterations.