Assessment and optimisation of normalisation methods for dual-colour antibody microarrays

Area of Science:

• Biotechnology
• Proteomics
• Bioinformatics

Background:

• Antibody microarray technology enables simultaneous protein expression measurement.
• Existing normalization methods from gene expression microarrays face challenges with protein arrays due to biased feature selection.
• Limited availability of high-quality antibodies and high costs lead to arrays focusing on regulated targets, underrepresenting housekeeping features.

Purpose of the Study:

• To evaluate and improve normalization methods for antibody-based protein microarrays.
• To address the violation of assumptions in traditional normalization techniques when applied to protein data.
• To enhance the accuracy of detecting differentially expressed proteins and improve classification power.

Main Methods:

• Comparison of established dual-colour gene expression microarray normalization methods.
• Development and application of an improved invariant selection algorithm for protein microarrays.
• Simulation studies to assess the impact of normalization on differential expression detection.
• Application to a pancreatic cancer dataset to evaluate classification performance.

Main Results:

• The improved invariant selection algorithms demonstrate superior performance compared to other normalization methods.
• These algorithms effectively address normalization biases inherent in protein microarrays.
• The methods show significant improvements in detecting differentially expressed features in simulation studies.

Conclusions:

• Improved invariant selection algorithms offer superior normalization for antibody microarrays.
• These methods are particularly effective when standard normalization assumptions are not met.
• The enhanced normalization improves the reliability of differential protein expression analysis and downstream applications like cancer classification.