When should one subtract background fluorescence in 2-color microarrays?

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Two-color microarrays are vital for genomic analysis but susceptible to noise, impacting gene expression inferences.
• Background fluorescence is a significant noise source, and its subtraction from spot fluorescence is a debated practice.
• Existing criteria for background subtraction are unclear, leading to inconsistent application in microarray data analysis.

Purpose of the Study:

• To systematically examine the bias-variance trade-off associated with background subtraction in two-color microarrays.
• To identify key factors influencing the decision to perform background subtraction.
• To develop evidence-based recommendations and diagnostic visualizations for background subtraction decisions.

Main Methods:

• Utilized simulation based on data from self-versus-self microarray experiments.
• Avoided distributional assumptions in the simulation model.
• Analyzed the bias-variance trade-off under diverse experimental conditions.

Main Results:

• Identified critical factors determining the necessity of background subtraction.
• Highlighted the importance of the correlation between foreground and background intensity ratios.
• Demonstrated how simulation can elucidate the bias-variance trade-off.

Conclusions:

• Provides a formal analysis of the bias-variance trade-off in microarray background subtraction.
• Offers practical guidance for researchers on optimizing gene expression data analysis.
• Recommends diagnostic visualizations to aid background subtraction decisions, improving data reliability.