Microarray Data Processing Techniques for Genome-Scale Network Inference from Large Public Repositories.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Standard microarray data pre-processing methods often filter out numerous genes for genome-scale network inference.
• Aggregating diverse experiments introduces technical and biological variability, limiting gene inclusion.

Purpose of the Study:

• To introduce a novel pre-processing pipeline for genome-scale gene network inference from large microarray datasets.
• To enhance the scope of downstream network construction by addressing data heterogeneity.

Main Methods:

• Developed a pre-processing pipeline specifically designed for large-scale microarray data.
• Partitioned microarray datasets into tissue- and process-specific categories based on biological relevance.
• Applied the pipeline to infer genome-scale networks for Arabidopsis thaliana using 11,760 Affymetrix ATH1 chips.

Main Results:

• The proposed pre-processing pipeline significantly extends the limits of network construction.
• Demonstrated effectiveness using two distinct network inference methods on a large Arabidopsis thaliana dataset.
• Successfully inferred genome-scale networks, highlighting the pipeline's utility.

Conclusions:

• Data partitioning by biological relevance is crucial for effective genome-scale network inference from diverse microarray sources.
• The developed pipeline offers a robust solution for analyzing large public microarray repositories.
• The pipeline and associated datasets are publicly available to facilitate further research.