Probe classification of on-off type DNA microarray images with a nonlinear matching measure

Area of Science:

• Biotechnology
• Bioinformatics
• Genomics

Background:

• DNA microarrays are crucial for genetic analysis and disease diagnostics.
• Accurate signal detection is essential for reliable probe classification in microarrays.
• Existing signal detection methods may have limitations in sensitivity and specificity.

Purpose of the Study:

• To introduce and evaluate a novel nonlinear matching measure, the counting measure, for signal detection in DNA microarrays.
• To compare the performance of the counting measure against established methods.
• To assess the impact of a maximum response search method on the counting measure's efficacy.

Main Methods:

• The counting measure was defined as the number of 'on' pixels within a probe spot area.
• A maximum response search method was incorporated, involving maximizing responses across various positions and template sizes.
• The counting measure was applied to classify hybridized or non-hybridized probes on an on-off type DNA microarray.
• Performance was evaluated using data from 2390 patient samples on a human papillomavirus (HPV) DNA chip, comparing against normalized covariance and median measures.

Main Results:

• The counting measure demonstrated superior performance compared to normalized covariance and median methods.
• The enhanced performance was observed irrespective of whether the maximum response search method was employed.
• Combining the counting measure with a positional search yielded the most favorable results.

Conclusions:

• The counting measure is a highly effective nonlinear matching measure for signal detection in DNA microarrays.
• The counting measure, particularly when integrated with positional search, offers improved accuracy for probe classification.
• This method shows significant promise for applications like human papillomavirus (HPV) detection using DNA chips.