Influence of batch effect correction methods on drug induced differential gene expression profiles

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Batch effects are often unaddressed in computational drug repositioning studies using gene expression signatures.
• The impact of batch effect removal methods on signature-based drug repositioning remains unclear.

Purpose of the Study:

• To evaluate how different batch effect correction methods influence computational drug repositioning results using microarray data.
• To compare the performance of various batch effect correction techniques in the context of drug repositioning.

Main Methods:

• Differential analyses were performed on the Connectivity Map (CMAP) database.
• Several batch effect correction methods were applied and compared.
• External validity was assessed using connectivity mapping between CMAP and the Library of Integrated Network-based Cellular Signatures (LINCS) database.
• Simulation studies were conducted to assess statistical power.

Main Results:

• Gene signatures varied in size depending on the correction method used; limited agreement was found between Latent Effect Adjustment after Primary Projection (LEAPP) and Linear Models for Microarray Data (limma).
• Drug repositioning reliability was poor for samples smaller than 40, regardless of batch correction.
• Batch effect correction methods improved results significantly for sample sizes over 40 compared to no correction.
• The limma method correcting for two principal components showed the best performance in simulation studies.

Conclusions:

• Batch effect correction methods critically affect differential gene expression analysis and subsequent drug repositioning, particularly with sufficient sample size.
• It is recommended to include two to three principal components as covariates in limma models for studies with adequate sample size (over 40 combined drug and control samples).