Harshlight

Harshlight detects and masks blemishes on high-density oligonucleotide array (HDONA) microarray chips to improve the accuracy of downstream gene expression analyses.

Key Features:

• Automatic detection and masking: Identifies and masks three distinct blemish types—localized defects affecting a few probes, diffuse defects impacting larger areas, and extended defects that may compromise entire chips.
• Statistical and image-processing methods: Employs a combination of statistical techniques and image processing methods to identify and categorize microarray artifacts.
• Correction mechanism: Masks defective areas by substituting values with 'NA' or with the median value from corresponding probes on other chips within the dataset and performs substitution as a whole matrix to manage memory and maintain data integrity.
• Improved data analysis with MAS5: When used with the MAS5 algorithm, restores approximately two-thirds of gene expression values altered by compact blemishes, which otherwise cause an average of eight gene expressions per chip to change by more than 50%, including up to twofold changes.

Scientific Applications:

• High-precision gene expression analysis: Improves reliability of differential expression studies performed on HDONA microarray datasets.
• Microarray quality control: Detects and corrects chip-level artifacts to increase reproducibility across microarray experiments.

Methodology:

Uses statistical techniques and image processing methods to detect and categorize localized, diffuse, and extended blemishes, then masks defects by replacing values with 'NA' or median probe values from other chips, performing substitutions on the whole data matrix to manage memory.