consICA: an R package for robust reference-free deconvolution of multi-omics data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Understanding cellular processes and disease requires deciphering molecular signals from omics data.
• Robust and reproducible algorithms are crucial for extracting these signals effectively.

Purpose of the Study:

• To introduce the R/Bioconductor package consICA, a novel tool for analyzing heterogeneous omics data.
• To enable data-driven deconvolution for extracting biologically relevant molecular signals.

Main Methods:

• Consensus Independent Component Analysis (ICA) is employed as the core deconvolution method.
• The package integrates features for patient stratification and multimodal data integration.
• Parallel computing is implemented for efficient analysis on multicore systems.

Main Results:

• consICA effectively separates biological signals from technical noise in omics data.
• Extracted features are suitable for patient stratification and understanding cellular composition.
• The package provides built-in tools for signal interpretation, including annotation and survival analysis.

Conclusions:

• consICA offers a reproducible and efficient solution for analyzing complex molecular profiles.
• The package has significant implications for advancing cancer research and precision medicine.
• It facilitates the extraction of meaningful biological insights from diverse omics datasets.