Feature intensity reconstruction of biological probe array

• US 8,934,689 B2
• Filed: 10/30/2012
• Issued: 01/13/2015
• Est. Priority Date: 06/27/2006
• Status: Active Grant

• Current Assignee: Affymetrix, Inc. (Thermo Fisher Scientific Incorporated)
• Original Assignee: Affymetrix, Inc. (Thermo Fisher Scientific Incorporated)
• Inventors: Hubbell, Earl A.
• Primary Examiner(s): Park, Chan
• Assistant Examiner(s): Thomas, Mia M
• Application Number:
  US13/663,673
• Publication Number:
  US 20130142408A1
• Time in Patent Office: 805 Days
• Field of Search: 382128-133, 382173-175, 382/199, 382282-284, 435/4, 435/6.1, 600/352, 600/447, 600/476, 600/407, 702/19, 702/20, 702/27, 702/47
• US Class Current: 382/129
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation
• CPC Class Codes: G06V 2201/04 Recognition of patterns in DNA microarrays
• CPC Class Codes: G16B 25/00 ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression

Abstract

The invention provides methods and systems for reconstructing feature intensities from pixel level data. In certain embodiments, the invention uses an empirically determined transfer function to construct a theoretical estimate of pixel level data and then iteratively updates feature intensities based on a minimum multiplicative error between the pixel level data and the theoretical estimate of the pixel level data.

Claims

1. A computer-implemented method of determining feature intensities from image data from a probe array, the method comprising:

a. obtaining image data for a plurality of features of the probe array, wherein the image data comprises a set of pixels, wherein the set of pixels comprises groups of pixels associated with each feature, wherein the groups are aligned with each feature, wherein the groups of pixels indicate a center pixel for each feature;

b. determining observed pixel intensities for one or more pixels in the set of pixels;

c. determining initial feature intensities for the plurality of features, wherein the initial feature intensities are based upon the observed pixel intensities;

d. determining a theoretical pixel intensity for one or more pixels in the set of pixels from feature intensities of one or more of the plurality of features;

e. determining an error for each pixel in the set of pixels, wherein the error is based upon the observed pixel intensity and the theoretical pixel intensity; and

f. determining a corrected feature intensity for the plurality of features, wherein the corrected feature intensity is based upon a previously determined feature intensity and the error, and

g. repeating steps d-f comparing the corrected feature intensities against previously determined feature intensities until the corrected feature intensities converge; and

h. after convergence, outputting the adjusted feature intensities.

2. The method of claim 1,

wherein the theoretical pixel intensities are based, at least in part, upon a plurality of feature intensities;

wherein determining the theoretical pixel intensity comprises determining an empirically based transfer function, and

wherein determining the theoretical pixel intensity comprises multiplying the empirically based transfer function with a plurality of feature intensities.

3. The method of claim 2, wherein the empirically based transfer function relates to a shape formed by of pixels associated with a feature.

4. The method of claim 2, wherein the empirically based transfer function relates for each pixel, the location of the pixel within the image data with feature intensities.

5. The method of claim 2, wherein the empirically based transfer function comprises an array, and wherein the array comprises dimensions relating to pixel locations and dimensions relating to an address of a plurality of features.

6. The method of claim 3, wherein the empirically based transfer function is based upon representation of a pixel location relative to a center location of a plurality of features.

7. The method of claim 2, wherein determining the theoretical pixel intensity additionally comprises determining a background intensity.

8. The method of claim 7, wherein the background intensity is added to an initial theoretical pixel intensity to determine the theoretical pixel intensity.

9. The method of claim 1, wherein determining the error comprises calculating a multiplicative error value based upon the observed pixel intensity and the theoretical pixel intensity.

10. The method of claim 9, wherein calculating the multiplicative error value comprises determining an error function based upon the observed pixel intensity and the theoretical pixel intensity.

11. The method of claim 10, wherein the error function is the quotient of dividing the observed pixel intensity by the theoretical pixel intensity.

12. The method of claim 10, wherein the error function is adjusted to remove outlying error values.

13. The method of claim 12, wherein the outlying error values are removed with use of a standard deviation function.

14. The method of claim 12, wherein determining the adjusted feature intensity comprises multiplying the initial feature intensity by an update factor, and wherein the update factor is based upon the error function after adjustment to remove outlying error values.

15. The method of claim 10, wherein the error function is adjusted through comparison to a preselected value.

16. The method of claim 15, wherein the error function is restricted within a preselected range, wherein the preselected range is defined by the preselected value and 1/preselected value.

17. The method of claim 1, wherein determining the adjusted feature intensity comprises multiplying the initial feature intensity by an update factor, and wherein the update factor is based upon the error.

18. The method of claim 17, wherein the update factor is additionally based upon a weight function, wherein each feature comprises a center location, and wherein the weight function assigns a weight to a pixel based upon the pixel'"'"'s distance from the center location such that pixels are assigned a higher weight as their distance from the center location decreases.

19. The method of claim 1, further comprising:

repeating the determining a theoretical pixel intensity, determining an error for each pixel in the set of pixels, and determining feature intensity steps until the adjusted feature intensity has converged, wherein the repeated third through fifth determining steps utilize the corrected feature intensity in place of the initial feature intensity.

20. The method of claim 1, wherein determining an observed pixel intensity step, the determining an initial feature intensity step, the determining a theoretical pixel intensity step, the determining an error for each pixel in the set of pixels step, and the determining an adjusted feature intensity step are performed by a computer processor located on a computer.