Estimation of expression levels in spotted microarrays with saturated pixels

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Laser scanning of spotted microarrays can produce saturated pixel values.
• Pixel saturation occurs when scan settings exceed the limit (L=65535), capping values at L.
• Uncorrected saturation leads to biased gene expression level estimations.

Purpose of the Study:

• To develop a method for imputing censored (saturated) pixel values in microarray images.
• To correct for bias introduced by pixel saturation in gene expression data.
• To enhance the dynamic range of detection in microarray experiments.

Main Methods:

• A linear model approach is proposed for imputing censored values.
• The model utilizes principal components derived from uncensored spots on the same array.
• The method is computationally efficient and adaptable to varying spot shapes and profiles.

Main Results:

• The proposed linear model effectively imputes censored pixel values.
• This method demonstrates superior performance compared to the polynomial-hyperbolic model in bias correction.
• Application to biological data shows enhanced dynamic range and improved accuracy.

Conclusions:

• The principal component-based linear model offers a robust solution for handling saturated pixels in microarray imaging.
• This approach improves the reliability of gene expression level estimates.
• The method has practical implications for increasing the sensitivity and accuracy of genomic analyses.