Analysing multiple types of molecular profiles simultaneously: connecting the needles in the haystack

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• A random-effects framework previously modeled gene expression associations with DNA copy number.
• This framework was extended to analyze mRNA and microRNA expression using gene sets.
• Previous methods focused on gene copy number and lacked integration with other molecular data.

Purpose of the Study:

• To extend existing models to incorporate multiple molecular profiles beyond DNA copy number.
• To investigate the impact of methylation changes and loss-of-heterozygosity (LOH) on gene expression.
• To develop a robust, genome-wide approach for identifying gene expression associations.

Main Methods:

• Extended a random-effects framework to include methylation and LOH data alongside copy number.
• Utilized gene-set modeling for improved robustness and statistical power.
• Applied the method to genome-wide datasets from colon and breast cancer samples.

Main Results:

• Identified gene expression-regulating mechanisms involving copy number, methylation, or both.
• Discovered distinct molecular mechanisms influencing gene expression in different samples.
• Demonstrated the method's ability to separate true associations from noise.

Conclusions:

• The developed method effectively analyzes associations between high-dimensional molecular datasets.
• It is computationally efficient, flexible, and powerful for genomic studies.
• The approach is applicable to various molecular data types and is available as a Bioconductor package.