Detection and correction of probe-level artefacts on microarrays

Area of Science:

• Bioinformatics
• Genomics
• Microarray Analysis

Background:

• Spatial defects are common in Affymetrix microarrays within the Gene Expression Omnibus (GEO) dataset.
• Current gene expression analysis pipelines rarely incorporate artefact detection.
• Existing methods for spatial noise detection and correction are insufficient, often relying on summarization or discarding arrays.

Purpose of the Study:

• To address the limitations of current methods in detecting and correcting spatial artefacts on microarrays.
• To develop a robust and easily integrated approach for identifying and rectifying defective probe measurements.

Main Methods:

• Developed a simple yet effective approach for detecting spatial artefacts with high recall and precision.
• Enhanced artefact detection by incorporating the spatial layout of microarrays.
• Proposed two correction methods: substituting defective probe values using probeset information and filtering corrupted probes.

Main Results:

• Demonstrated that standard robust summarization procedures are vulnerable to array artefacts and do not correct them effectively.
• Showcased the developed approach's ability to accurately identify and correct defective probe measurements.
• Confirmed that the proposed methods outperform existing tools in artefact correction.

Conclusions:

• Summarization methods are inadequate for correcting defective probes; the presented identification and correction methods offer a straightforward solution.
• The developed methods generate corrected CEL files, enabling seamless integration into existing microarray analysis pipelines as a pre-processing step.
• An R package for these artefact correction methods is available for free use.