Method and system for dynamic, automated detection of outlying feature and feature background regions during processing of data scanned from a chemical array

• US 20080089568A1
• Filed: 10/12/2006
• Published: 04/17/2008
• Est. Priority Date: 10/12/2006
• Status: Active Grant

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Agilent Technologies, Inc.
• Inventors: Delenstarr, Glenda C.
• Granted Patent: US 8,077,951 B2
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G01N 21/253 for batch operation, i.e. multisample apparatus analytical automats G01N35/00
• CPC Class Codes: G06T 2207/30072 Microarray; Biochip, DNA array; Well plate
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation

Abstract

A method and system for employing signal-intensity data contained within areas of a scanned image of a chemical array corresponding to features and feature backgrounds in order to determine whether or not the features or feature backgrounds have non-uniform signal intensities and are thus outlier features and outlier feature backgrounds. A signal intensity variance model is provided to dynamically adjust to the signals read from the scanned image of the array. Calculated variance from measured values of the signal intensities and pixel distributions within a feature or feature background is compared to a maximum allowable variance calculated for the feature or feature background based on the dynamically adjusting signal intensity variance model. When the calculated variance (from the measured values) is less than or equal to the maximum allowable variance, the feature or feature background is considered to have acceptable signal-intensity uniformity. Otherwise, the feature or feature background is flagged as an outlier feature or outlier feature background.

Claims

1. A method of identifying a non-uniform measured signal distribution in a region of a scanned image of a chemical array, said method comprising:

providing an automatic, dynamically adjustable variance model for measured signal distributions within regions of the chemical array;

determining a variance of measured signals within the region;

determining whether or not the region contains a non-uniform measured signal distribution by comparing the determined variance of measured signals within the region to the variance model; and

displaying results indicating whether or not the region contains a non-uniform measure signal distribution.

2. The method of claim 1, wherein the dynamic automatically adjustable variance model includes coefficients of the variance model that are dependent upon a scanner used to scan the image and coefficients that are dependent upon the characteristics of the chemical array, wherein the coefficients that are dependent upon the characteristics of the chemical array are dynamically determined by the model.

3. The method of claim 1 further including determining a variance threshold from the automatic, dynamically adjustable variance model, and wherein comparing the determined variance of measured signals within the region to the variance model comprises comparing the determined variance of measured signals within the region to the determined variance threshold.

4. The method of claim 3, wherein the scanned image comprises pixels, each pixel associated with a count representing a signal measured from a corresponding portion of the chemical array.

5. The method of claim 4, wherein the variance model is a linear combination of model variance terms.

6. The method of claim 5, wherein said model variance terms include:

a variance term arising from non-uniformities associated with target-molecule labeling, feature synthesis, probe molecule application, and other solution and surface chemistry effects;

an automatic, dynamically adjusting variance term arising from non-uniformities associated with scanner counting errors, that automatically and dynamically adjusts to the pixels of the scanned image; and

an automatic, dynamically adjusting variance term arising from non-uniformities associated with electronic noise in a scanner, background-level signal noise arising from a chemical array substrate, and other noise, that automatically and dynamically adjusts to the pixels of the scanned image.

7. The method of claim 6, wherein said variance term arising from non-uniformities associated with target-molecule labeling, feature synthesis, probe molecule application, and other solution and surface chemistry effects is an automatic, dynamically adjusting variable that adjusts to the pixels of the scanned image.

8. The method of claim 6, wherein non-uniformities associated with scanner counting errors are modeled by a Poisson distribution.

9. The method of claim 4, wherein the variance model is an expression including a mean pixel count for the region as a variable.

10. The method of claim 3, wherein said calculating a variance threshold from the variance model further includes assuming a chi-squared distribution for one less than the number of pixels multiplied by the model variance and divided by the theoretical variance of measured signals within the region, and, based on the chi-squared distribution assumption, selecting a threshold variance value below which the determined variance of measured signals within the region has a high probability of indicating an acceptably uniform distribution of measured signals within the region.

11. The method of claim 1, wherein the region is selected from among a feature and a feature background.

12. The method of claim 1, wherein the variance model is provided according to chemical and physical properties of the chemical array, electronic and physical properties of a scanning device, and experimental conditions to which the chemical array is exposed.

13. A computer-readable medium having computer executable instructions stored thereon, for performing the method of claim 1.

14. A method of dynamically tuning a variance model to an instance of a hybridized chemical array, said method comprising the steps of:

inputting signal intensity values produced from scanning a region of the hybridized array;

defining a variance model as a sum of factors multiplied by multiplicative coefficients, wherein the multiplicative coefficients for the respective factors are each a composite of a scanner-dependent multiplier and an array-dependent multiplier;

dynamically calculating the array-dependent multipliers;

determining a variance of measured signals within the region;

determining whether or not the region contains a non-uniform measured signal distribution by comparing the determined variance of measured signals within the region to the variance model; and

displaying results indicating whether or not the region contains a non-uniform measure signal distribution.

15. The method of claim 14, wherein the scanner-dependent multipliers are tuned from empirical analysis of large numbers of arrays having different characteristics, such that the scanner-dependent multipliers stay the same in the variance model as long as the same scanner or model of scanner is being used with various arrays, and wherein the array-dependent multipliers are dynamically calculated for each individual array that the variance model is applied to.

16. A system for identifying a non-uniform measured signal distribution in a region of a scanned image of a chemical array, the system comprising:

a digital representation of the measured signals in the region of the scanned image of the chemical array stored within a memory component;

an automatic, dynamically adjustable variance model for measured signal distributions within regions of the chemical array stored within a memory component; and

a computational processing engine that calculates a variance of measured signals within the region and compares the calculated variance with the variance model to determine whether or not the region contains a non-uniform measured signal distribution by comparing the determined variance of measured signals within the region to the variance model.

17. The system of claim 16, wherein the automatic, dynamically adjusting variance model further includes a variance threshold to which the computational processing engine compares the calculated variance.

18. The system of claim 16, wherein the digital representation of the measured signals in the region of the scanned image of the chemical array comprises a number of pixels, each pixel associated with a count representing a signal measured from a corresponding portion of the chemical array.

19. The system of claim 16, wherein the automatic, dynamically adjusting variance model is a linear combination of model variance terms.

20. The system of claim 19, wherein said model variance terms include:

a variance term arising from non-uniformities associated with target-molecule labeling, feature synthesis, probe molecule application, and other solution and surface chemistry effects;

an automatic, dynamically adjusting variance term arising from non-uniformities associated with scanner counting errors, that automatically and dynamically adjusts to the pixels of the scanned image; and

an automatic, dynamically adjusting variance term arising from non-uniformities associated with electronic noise in a scanner, background-level signal noise arising from a chemical array substrate, and other noise, that automatically and dynamically adjusts to the pixels of the scanned image.

21. The system of claim 20, wherein said variance term arising from non-uniformities associated with target-molecule labeling, feature synthesis, probe molecule application, and other solution and surface chemistry effects is an automatic, dynamically adjusting variable that adjusts to the pixels of the scanned image.

22. The system of claim 20, wherein non-uniformities associated with scanner counting errors are modeled by a Poisson distribution.

23. The system of claim 16, wherein the variance model is an expression including a mean pixel count for the region as a variable.

24. The system of claim 16, wherein the variance model is based on chemical and physical properties of the chemical array, electronic and physical properties of a scanning device, and experimental conditions to which the chemical array is exposed.