Identification of population-level differentially expressed genes in one-phenotype data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Analyzing gene expression in tissues like the heart and brain is challenging due to the difficulty in obtaining normal controls.
• Standard methods (e.g., SAM, edgeR, limma) require case-control samples and cannot analyze one-phenotype data effectively.
• The RankComp algorithm identifies individual-level differentially expressed genes (DEGs) but lacks dysregulation direction information for population-level analysis.

Purpose of the Study:

• To optimize the RankComp algorithm for analyzing one-phenotype gene expression data.
• To develop a method that identifies both differentially expressed genes (DEGs) and their dysregulation directions from limited sample types.
• To improve the detection power and robustness of DEG analysis for datasets lacking normal controls.

Main Methods:

• Optimization of the RankComp algorithm, resulting in a new tool named PhenoComp.
• Application of PhenoComp to simulated and real one-phenotype gene expression datasets.
• Comparison of PhenoComp's performance against common DEG analysis methods using case-control data as a gold standard.

Main Results:

• PhenoComp successfully identifies differentially expressed genes (DEGs) and their dysregulation directions in one-phenotype data.
• PhenoComp demonstrates robust detection power, outperforming the original RankComp algorithm.
• DEGs identified by PhenoComp using one-phenotype data are comparable in accuracy to those found by standard methods using case-control samples.
• PhenoComp performs well even with weak differential expression signals.

Conclusions:

• PhenoComp provides a powerful and robust solution for analyzing gene expression data with limited sample types (one phenotype).
• The algorithm accurately identifies DEGs and their direction of change, overcoming limitations of previous methods.
• PhenoComp offers a valuable alternative for studying complex tissues like the heart and brain where normal controls are scarce.