MDQC: a new quality assessment method for microarrays based on quality control reports

Area of Science:

• Genomics
• Bioinformatics
• Data Science

Background:

• Microarray data production involves multiple steps, increasing the risk of technical issues that compromise data quality.
• Identifying low-quality arrays is essential for reliable downstream analysis.
• Existing quality control methods may not fully capture complex data quality issues.

Purpose of the Study:

• To develop and evaluate a novel multivariate approach for assessing microarray data quality.
• To identify potential sources of quality problems in microarray datasets.
• To provide a computationally inexpensive and interpretable method for quality control.

Main Methods:

• Proposed Mahalanobis Distance Quality Control (MDQC), a multivariate approach for outlier detection.
• Examined different variations of the MDQC method.
• Applied MDQC to analyze quality attributes of microarray arrays using their quality control (QC) reports.
• Utilized case studies to validate the effectiveness of the multivariate approach.

Main Results:

• MDQC effectively flags arrays with quality attributes deviating from the norm.
• Multivariate analysis provides richer insights compared to univariate parameter examination.
• The method is computationally inexpensive and results are easily visualized and interpreted.
• Analyzing subsets of quality measures enhances detection of unusual arrays and aids in identifying problem sources.

Conclusions:

• MDQC offers a robust and efficient method for microarray data quality assessment.
• The approach aids in identifying arrays with quality issues and their potential causes.
• A Bioconductor library for MDQC implementation will be available, facilitating its adoption.