Provided with the DNA microarray scanner for use with Agilent microarrays, Agilent's sophisticated Feature Extraction and Analysis Software identifies which specific genes are differentially expressed in the form of log ratios. In addition to providing basic statistical data, the software provides an error value and a p-value for the log ratio of each feature, which enables researchers to evaluate the confidence in their data for each feature. Agilent’s Feature Extraction software has added functionality to analyze images and extract data from a wide mix of both Agilent and non-Agilent 1” x 3” microarrays scanned on the Agilent Microarray Scanner.

Features

High Quality Data Analysis Achieved Through ...

• Feature Identification - Feature pixels and background pixels are automatically identified and marked for each feature on the array.
• Inlier/Outlier Analysis - All pixels in a feature and background region are scrutinized to identify pixels with signal intensity outside of a given statistical range. Outlier pixels are removed.
• Feature Flagging - Flagging algorithms are applied to every feature and its background region to compensate for imperfections caused by bubbles, scratches, bright pixel contaminants and abnormalities. Flagged features are identified but not discarded from the analysis.
• Automatic gradient detection – Detect and automatically remove spatial intensity gradients that can bias the result of microarray analysis.
• Background Subtraction - Offset signal from the scanner and background signal from the non-specific binding of label contibute to feature signal on the array. Negative control probes or other background subtraction techniques can be used to estimate and compensate for global background. Alternatively, the local background can be used to calculate and subtract the background signal.
• Cross Hybridization Flagging - A second method can be used to control non-specific hybridization on Agilent microarrays. For each perfect complement probe on Agilent in-situ microarrays, a companion deletion probe with one base missing from the middle of the probe can be included. If the signal from the perfect match complement is not significantly higher than the deletion probe signal, the feature is flagged. Dye Normalization - Label bias effects result in the appearance of differential expression where none actually exists. Dye normalization algorithms are used to force the average log ratio between the red and green signals to be zero for non-differentially expressed genes.
• Statistical Analysis - For each feature and background region, both a red and a green signal pixel error are calculated from the population of pixels that define the feature. This error provides an estimate of the signal error of a feature, and is propogated through each step of the Feature Extraction algorithm. Additionally, the software propagates the background error and, if applicable, the deletion control pixel error. Once the total log ratio error has been determined, a statistical confidence value (p-value) is calculated. The confidence value enables the experimenter to evaluate whether the feature log ratio is significantly differentially expressed based upon the confidence value set by the user.
• Data Output - The extracted file output is exported in a text (tab-delimited) format or an XML-based format that is tailored for use with Rosetta Resolver? Bioinformatics Software. The use of multiple flagging techniques can identify a high quality sub-population of array features with reduced noise at all signal intensities. The flexibility of background subtraction and adjustable automatic outlier flagging enable customized data analysis that can be pre-configured to consistent, streamlined data analysis that you can be confident in every time.