Genome-Wide Uncertainty-Moderated Extraction of Signal Annotations from Multi-Sample Functional Genomics Data

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Sequencing-based functional genomics datasets are often corrupted by noise.
• Accurate signal estimation is crucial for downstream analyses like differential analysis and identifying functional genomic regions.

Purpose of the Study:

• To present Consenrich, a novel technique for genome-wide signal estimation in noisy multi-sample sequencing data.
• To improve the accuracy of signal extraction from functional genomics datasets.

Main Methods:

• Consenrich employs a sequential prediction-correction framework.
• It models spatial dependencies between genomic loci and regional, sample-specific noise processes.

Main Results:

• Consenrich demonstrates significant improvements over existing benchmarks in challenging estimation problems.
• The technique effectively reconciles noisy functional genomics data samples.

Conclusions:

• Consenrich offers a principled and effective method for extracting reliable signals from noisy sequencing data.
• The refined signal extraction has practical applications in differential analyses and identifying functionally enriched genomic regions.