Microarray performance management system

• US 20040019466A1
• Filed: 04/23/2003
• Published: 01/29/2004
• Est. Priority Date: 04/23/2002
• Status: Abandoned Application

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Agilent Technologies, Inc.
• Inventors: Watson, George A., Dickinson, Robert D., Illouz, Mika, Minor, James M.
• Application Number:
  US10/422,570
• Publication Number:
  US 20040019466A1
• Time in Patent Office: Days
• US Class Current: 702/190
• CPC Class Codes: G16B 25/00 ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression

Abstract

Systems, methods and computer readable media for evaluating profiles of multi-variable data values. Systems, methods and computer readable media for removing nuisance distortions from data to provide “cleaner”, more reliable data.

Claims

1. A method of removing nuisance distortions from array data, said method comprising the steps of:

inputting signals generated from reading an array containing biological data;

applying at least one algorithm to the inputted signals to estimate and remove at least one nuisance distortion from the signals representing the array data; and

outputting signals representative of the biological data after removal of the at least one nuisance distortion.

2. The method of claim 1, wherein the biological data comprises biological information or content of interest in an experiment, and wherein signals representative of the biological data are retained in the outputted signals.

3. The method of claim 1, wherein said applying at least one algorithm to estimate and remove comprises de-trending.

4. A method comprising forwarding a result obtained from the method of claim 1 to a remote location.

5. A method comprising transmitting data representing a result obtained from the method of claim 1 to a remote location.

6. A method comprising receiving a result obtained from a method of claim 1 from a remote location.

7. The method of claim 1, wherein said applying at least one algorithm to estimate and remove comprises gradient analysis and filtering of the inputted signals.

8. The method of claim 1, wherein said applying at least one algorithm to estimate and remove includes entering at least one blocking factor to remove blocking effects.

9. The method of claim 8, further comprising employing generalized response-surface methodology to approximate global gradient effects of the inputted data.

10. The method of claim 3, wherein said de-trending removes all low frequency and block patterns from the inputted data.

11. The method of claim 3, wherein said de-trending is carried out by a harmonic model plus shift effects.

12. The method of claim 3, wherein said de-trending comprises use of a second-order polynomial plus shift factors.

13. The method of claim 12, wherein said de-trending further comprises use of a statistical predictor method using similarity transforms.

14. The method of claim 3, further comprising optimizing said de-trending by least squares regression.

15. The method of claim 1, wherein said non-biological distortions are selected from at least one of the group consisting of array patterns, channel bias and build bias.

16. The method of claim 1, wherein said non biological distortions include array patterns, channel bias and build bias.

17. The method of claim 1, wherein said inputting signals comprises inputting signals through at least two channels, each channel inputting a distinct set of signals generated from reading an array containing biological data.

18. The method of claim 17, further comprising comparing outputted results from the at least two channels;

and de-warping the outputted results based on error properties of the outputted results.

19. A method comprising forwarding a result obtained from the method of claim 18 to a remote location.

20. A method comprising transmitting data representing a result obtained from the method of claim 18 to a remote location.

21. A method comprising receiving a result obtained from a method of claim 18 from a remote location.

22. The method of claim 1, further comprising assessing the analytical performance of the array using at least one of the metrics selected from the group consisting of accuracy, precision, standard deviation of signal response, coefficient of variation of signal response, comparative precision, concordance correlation, minimum detectable fold change, analytical sensitivity, threshold, limit of detectable response, limit of detection, limit of quantifiable signal response, limit of quantitation, linearity, dynamic range, linear dynamic range and signal response range.

23. The method of claim 1, further comprising assessing the quality of the array using at least one of the metrics selected from the group consisting of statistical calibration, asymmetry in signal response, mean signal response, standard deviation in signal response, coefficient of variation in signal response, skewness, kurtosis and control charting.

24. A computer-implemented method of evaluating profiles of multi-variable data values, said method comprising the steps of:

training on a training set of profiles having been previously evaluated;

determining a computer functional relationship between the training profiles and values assigned to the respective training profiles during the previous evaluation; and

applying the computer functional relationship to one or more profiles not belonging to the training set to generate evaluation values for the one or more profiles not belonging to the training set.

25. The method of claim 24, wherein the training set of profiles have been previously evaluated by one or more human inspectors.

26. A method comprising forwarding a result obtained from the method of claim 24 to a remote location.

27. A method comprising transmitting data representing a result obtained from the method of claim 24 to a remote location.

28. A method comprising receiving a result obtained from a method of claim 24 from a remote location.

29. The method of claim 24, wherein each of said profiles of said training set and each of said profiles not belonging to the training set comprises data characterizing a microarray.

30. The method of claim 24, wherein said multi-variable data values comprise metrics generated from analyzing signals generated from reading a microarray.

31. The method of claim 30, wherein said metrics comprise at least one of an estimated parameter from de-trending, an estimated parameter from de-warping, a delta metric, an accuracy metric, a precision metric, an analytical sensitivity metric, a linearity metric, a dynamic and linear dynamic range metric or a statistical calibration.

32. The method of claim 24, wherein said training comprises de-trending the training set of profiles.

33. The method of claim 32, wherein said training further comprises de-warping the training set of profiles.

34. The method of claim 32, wherein said training further comprises creating metrics representative of the de-trended profiles.

35. The method of claim 33, wherein said training further comprises creating metrics representative of the de-trended profiles after said de-trending, and creating metrics representative of the de-warped profiles after said de-warping

36. The method of claim 24, wherein said determining a computer function relationship comprises application of an initial model and development of a more complex model to determine the computer functional relationship.

37. The method of claim 36, wherein the initial model comprises Model Zero.

38. The method of claim 36, wherein the more complex model comprises a predictor-corrector model.

39. The method of claim 24, wherein said evaluation values comprises classification and quality scoring values.

40. The method of claim 39, wherein said classification and quality scoring values comprise pass, fail and marginal.

41. A method of automatically classifying and quality scoring microarrays containing biological data, said method comprising the steps of:

training on a training set of microarrays having been previously classified and scored by one or more human inspectors;

determining a computer functional relationship between the training set and classification and scoring values assigned to the respective training microarrays in the training set by the one or more human inspectors; and

applying the computer functional relationship to one or more microarrays containing biological data and not belonging to the training set, and automatically classifying and scoring the one or more microarrays not belonging to the training set, based on the computer functional relationship.

42. The method of claim 41, wherein said training comprises:

de-trending each microarray in the training set;

creating metrics representative of each microarray after de-trending;

de-warping each microarray in the training set;

creating metrics representative of each microarray after de-warping; and

generating a profile for each microarray in the training set, wherein each profile contains said metrics after de-trending and de-warping; and

wherein said determining a computer functional relationship comprises;

applying an initial model and developing a more complex model to determine the computer functional relationship between said profiles and said classification and quality scores having been previously assigned to said training set.

43. The method of claim 41, wherein said applying the computer functional relationship to one or more microarrays containing biological data and note belonging to the training set, and automatically classifying and scoring the microarrays not belonging to the training set, based on the computer functional relationship comprises:

de-trending each microarray not belonging to the training set;

creating metrics representative of each microarray not belonging to the training set, after de-trending;

de-warping each microarray not belonging to the training set;

creating metrics representative of each microarray not belonging to the training set, after de-warping; and

generating a profile for each microarray not belonging to the training set, wherein each profile contains said metrics after de-trending and de-warping; and

applying said computer functional relationship to said profiles characterizing said microarrays not belonging to the training set, to automatically classify and quality score said microarrays not belonging to the training set.

44. A system for removing nuisance distortions from array data, said system comprising:

means for inputting signals generated from reading an array containing biological data;

means for applying at least one algorithm to the inputted signals to estimate and remove at least one nuisance distortion from the signals representing the array data; and

means for outputting signals representative of the biological data after removal of said at least one nuisance distortion.

45. A computer readable medium carrying one or more sequences of instructions for removing nuisance distortions from array data, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

inputting signals generated from reading an array containing biological data;

applying at least one algorithm to the inputted signals to estimate and remove at least one nuisance distortion from the signals representing the array data; and

outputting signals representative of the biological data after removal of said at least one nuisance distortion.

46. A system for evaluating profiles of multi-variable data values, said system comprising:

means for training on a training set of profiles having been previously evaluated;

means for determining a computer functional relationship between the training profiles and values assigned to the respective training profiles during the previous evaluation; and

means for applying the computer functional relationship to one or more profiles not belonging to the training set to generate evaluation values for the one or more profiles not belonging to the training set.

47. A computer readable medium carrying one or more sequences of instructions for evaluating profiles of multi-variable data values, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

training on a training set of profiles having been previously evaluated;

determining a computer functional relationship between the training profiles and values assigned to the respective training profiles during the previous evaluation; and

applying the computer functional relationship to one or more profiles not belonging to the training set to generate evaluation values for the one or more profiles not belonging to the training set.

48. A system for automatically classifying and quality scoring microarrays containing biological data, said system comprising:

means for training on a training set of microarrays having been previously classified and scored;

means for determining a computer functional relationship between the training set and classification and scoring values assigned to the respective training microarrays in the training set; and

means for applying the computer functional relationship to one or more microarrays containing biological data and not belonging to the training set, and automatically classifying and scoring the microarrays not belonging to the training set, based on the computer functional relationship.

49. A computer readable medium carrying one or more sequences of instructions for automatically classifying and quality scoring microarrays, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

training on a training set of microarrays having been previously classified and scored;

determining a computer functional relationship between the training set and classification and scoring values assigned to the respective training microarrays in the training set; and

applying the computer functional relationship to one or more microarrays containing biological data and not belonging to the training set, and automatically classifying and scoring the microarrays not belonging to the training set, based on the computer functional relationship.