A ratiometric-based measure of gene co-expression

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Traditional gene co-expression analysis relies on correlation and mutual information, focusing on similar expression trajectories.
• These methods may miss biologically relevant gene relationships characterized by stable ratios or constrained variance.
• Homogeneous sample sets and single-cell RNA sequencing present unique contexts where conventional methods might be insensitive.

Purpose of the Study:

• To develop and implement a novel ratiometric approach for detecting gene associations (RA).
• To evaluate the effectiveness of RA in identifying gene relationships missed by conventional methods.
• To assess the applicability of RA to both bulk and single-cell RNA sequencing data.

Main Methods:

• Developed a ratiometric approach (RA) based on the coefficient of variation of gene expression ratios.
• Applied RA to lymphoblastoid RNA-sequencing data from the 1000 Genomes Project Consortium.
• Analyzed single-cell RNA sequencing data from individual lymphoblastoid cells.

Main Results:

• The RA method identified approximately 1/4 of all possible gene pairs in the homogeneous lymphoblastoid dataset.
• RA successfully captured known biological relationships and revealed novel associations missed by correlation and mutual information methods.
• The desirable properties of RA were confirmed in single-cell RNA sequencing data.

Conclusions:

• The ratiometric method effectively identifies biologically significant gene relationships in homogeneous datasets, often overlooked by conventional approaches.
• RA's ability to detect these relationships opens new avenues for investigating gene regulatory mechanisms.
• RA is a scalable method well-suited for analyzing large transcriptomic datasets, including bulk and single-cell RNA-seq.