Characterizing biological stimuli by response curves

• US 20050137806A1
• Filed: 07/16/2004
• Published: 06/23/2005
• Est. Priority Date: 07/18/2003
• Status: Active Grant

• Current Assignee: Cytokinetics, Inc.
• Original Assignee: Cytokinetics, Inc.
• Inventors: Vaisberg, Eugeni A., Coleman, Daniel A., Kutsyy, Vadim
• Granted Patent: US 7,246,012 B2
• Time in Patent Office: Days
• US Class Current: 702/19
• CPC Class Codes: G06F 18/24143 Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN]
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06V 20/698 Matching; Classification

Abstract

A method for calculating distances between stimulus response curves (e.g., dose response curves) allows classification of stimuli. The response curves show how the phenotype of one or more cells changes in response to varying levels of the stimulus. Each “point” on the curve represents quantitative phenotype or signature for cell(s) at a particular level of stimulus (e.g., dose of a therapeutic). The signatures are multivariate phenotypic representations of the cell(s). They include various features of the cell(s) obtained by image analysis. To facilitate the comparison of stimuli, distances between points on the response curves are calculated. First, the response curves may be aligned on a coordinate representing a separate distance, r, from a common point of negative control (e.g., the point where no stimulus is applied). Integration on r may be used to compute the distance between two response curves. The distance between response curves is used to classify stimuli.

Claims

1. A method of classifying a stimulus based on its associated stimulus response path in biological feature space, the method comprising:

(a) aligning two or more stimulus response paths on a common radius, r, from a point in biological feature space where a negative control is applied;

(b) calculating a separation distance, D, between points on two of the stimulus response paths having the common radius, r, from the negative control point; and

(c) classifying the stimulus of at least one of the stimulus response paths on the basis of the calculated value of the separation distance, D.

2. The method of claim 1, wherein aligning the two or more stimulus response paths comprises calculating the common radius, r, as a Euclidean distance from the negative control point.

3. The method of claim 1, wherein calculating the separation distance, D, comprises calculating a value based on a spherical geometry including the points sharing the common radius, r, from the negative control point.

4. The method of claim 3, wherein the value based on a spherical geometry comprises an angle or an inner product.

5. The method of claim 1, wherein the stimulus response paths are associated with particular mechanisms of action, and wherein (c) comprises classifying the at least one of the stimulus response paths based on the mechanism of action.

6. The method of claim 1, wherein the stimulus response paths comprise points that define signatures, and wherein the signatures comprise phenotype features of a biological system exposed to stimuli.

7. The method of claim 6, wherein the signatures are obtained by applying varying levels of the stimuli to the biological system.

8. The method of claim 6, wherein the signatures are obtained by monitoring the effect of the stimulus on the biological system over time.

9. The method of claim 6, wherein the biological system is a cell line.

10. The method of claim 1, wherein the stimulus is exposure to a chemical compound.

11. A method of classifying stimuli based on their effects on biological systems, the method comprising:

(a) receiving one or more signatures of a stimulus on one or more biological systems, wherein the signatures comprise multivariate data obtained from image analysis of biological systems exposed to the stimulus at specified levels of exposure to the stimulus;

(b) computing distances between (i) at least one of the signatures or points on a stimulus response path through multiple signatures and (ii) at least two regions of biological feature space associated with defined stimulus classifications; and

(c) classifying the stimulus based on the distances computed in (b).

12. The method of claim 11, wherein computing the distances in (b) comprises computing an angle or an inner products of vectors, each from a common center point in the biological feature space, wherein a first vector passes through one of the signatures or points of the stimulus and a second vector passes through a point on a stimulus response path associated with a classification in at least one of the regions of biological feature space.

13. The method of claim 11, wherein computing the distances in (b) comprises computing Euclidean distances.

14. The method of claim 11, wherein computing the distances in (b) comprises using a nearest neighbor computation.

15. The method of claim 11, wherein the regions of biological feature space are associated with particular mechanisms of action, and wherein (c) comprises classifying the at least one of the stimulus response paths based on the mechanism of action.

16. The method of claim 11, wherein the signatures are obtained by applying varying levels of the stimulus to the biological system.

17. The method of claim 11, wherein the signatures are obtained by monitoring the effect of the stimulus on the biological system over time.

18. The method of claim 11, wherein the biological system is a cell line.

19. The method of claim 11, wherein the stimulus is exposure to a chemical compound.

20. A computer program product comprising a machine readable medium on which is provided program instructions for classifying a stimulus based on its associated stimulus response path in biological feature space, the instuctions comprising:

(a) aligning two or more stimulus response paths on a common radius, r, from a point in biological feature space where a negative control is applied;

(b) calculating a separation distance, D, between points on two of the stimulus response paths having the common radius, r, from the negative control point; and

(c) classifying the stimulus of at least one of the stimulus response paths on the basis of the calculated value of the separation distance, D.

21. The computer program product of claim 20, wherein instructions for aligning the two or more stimulus response paths comprise instructions for calculating the common radius, r, as a Euclidean distance from the negative control point.

22. The computer program product of claim 20, wherein instructions for calculating the separation distance, D, comprise instructions for calculating a value based on a spherical geometry including the points sharing the common radius, r, from the negative control point.

23. The computer program product of claim 22, wherein the value based on a spherical geometry comprises an angle or an inner product.

24. The computer program product of claim 20, wherein the stimulus response paths are associated with particular mechanisms of action, and wherein (c) comprises instructions for classifying the at least one of the stimulus response paths based on the mechanism of action.

25. The computer program product of claim 20, wherein the stimulus response paths comprise points that define signatures, and wherein the signatures comprise phenotype features of a biological system exposed to stimuli.

26. An apparatus for classifying a stimulus based on its associated stimulus response path in biological feature space, the apparatus comprising:

(a) one or more processors; and

(b) memory,

wherein the one or more processors are designed or configured to classify the stimulus based on a separation distance, D calculated between points on stimulus response paths, which points are aligned on a common radius, r, from a point in biological feature space where a negative control is applied.

27. A computer program product comprising a machine readable medium on which is provided program instructions for classifying stimuli based on their effects on biological systems, the instructions comprising:

(a) receiving one or more signatures of a stimulus on one or more biological systems, wherein the signatures comprise multivariate data obtained from image analysis of biological systems exposed to the stimulus at specified levels of exposure to the stimulus;

(b) computing distances between (i) at least one of the signatures or points on a stimulus response path through multiple signatures and (ii) at least two regions of biological feature space associated with defined stimulus classifications; and

(c) classifying the stimulus based on the distances computed in (b).

28. The computer program product of claim 27, wherein the instructions for computing the distances in (b) comprise instructions for computing an angle or an inner products of vectors, each from a common center point in the biological feature space, wherein a first vector passes through one of the signatures or points of the stimulus and a second vector passes through a point on a stimulus response path associated with a classification in at least one of the regions of biological feature space.

29. The computer program product of claim 27, wherein the instructions for computing the distances in (b) comprise instructions for computing Euclidean distances.

30. The computer program product of claim 27, wherein the instructions for computing the distances in (b) comprise instructions for using a nearest neighbor computation.

31. The method of claim 27, wherein the regions of biological feature space are associated with particular mechanisms of action, and wherein the instructions in (c) comprise instructions for classifying the at least one of the stimulus response paths based on the mechanism of action.

32. An apparatus for classifying stimuli based on their effects on biological systems, the apparatus comprising:

(a) one or more processors; and

(b) a memory,

wherein the one or more processors are designed or configured to classify the stimulus based on the distances between (i) signatures or points on a stimulus response path through multiple signatures and (ii) regions of biological feature space associated with defined stimulus classifications, and

wherein the signatures comprise multivariate data obtained from image analysis of biological systems exposed to the stimulus at specified levels of exposure to the stimulus.

33. A method of creating a biological classification model for classifying the effect of stimuli on biological systems, the method comprising

(a) receiving a training set of stimulus response paths for stimuli having known classifications, wherein the stimulus response paths comprise signatures representing the multivariate response of stimuli at various levels;

(b) selecting a collection of biological features to be used in a proposed model;

(c) computing distances between the stimulus response paths in a biological feature space defined by the biological features selected in (b);

(d) characterizing the proposed model based on how well it groups stimulus response paths into the known classifications of the associated stimuli in the biological features space;

(e) repeating (b)-(d) for a plurality of selected collections of biological features; and

(f) choosing a proposed model as the biological classification model based on the characterizations made in (d).

34. The method of claim 33, wherein the signatures of a response path are obtained by applying varying levels of a stimulus to a biological system.

35. The method of claim 33, wherein the biological systems are a cell line.

36. The method of claim 33, wherein the stimuli comprise exposure to chemical compounds.

37. The method of claim 33, wherein the collection of biological features selected in (b) comprise one or more of morphological details, texture measures for a marker, intensity measures for a marker, statistical details, and values derived from any of the foregoing of a cell or cell population.

38. The method of claim 33, wherein computing the distances in (c) comprises computing an angle or an inner products of vectors, each from a common center point in the biological feature space, wherein a first vector passes through one of the signatures or points of a first stimulus response path and a second vector passes through one of the signatures or points of a second stimulus response path.

39. The method of claim 33, wherein computing the distances in (c) comprises computing Euclidean distances.

40. A computer program product comprising a machine readable medium on which is provided program instructions for creating a biological classification model for classifying the effect of stimuli on biological systems, the instructions comprising

(a) receiving a training set of stimulus response paths for stimuli having known classifications, wherein the stimulus response paths comprise signatures representing the multivariate response of stimuli at various levels;

(b) selecting a collection of biological features to be used in a proposed model;

(c) computing distances between the stimulus response paths in a biological feature space defined by the biological features selected in (b);

(d) characterizing the proposed model based on how well it groups stimulus response paths into the known classifications of the associated stimuli in the biological features space;

(e) repeating (b)-(d) for a plurality of selected collections of biological features; and

(f) choosing a proposed model as the biological classification model based on the characterizations made in (d).

41. The computer program product of claim 40, wherein the collection of biological features selected by the instructions in (b) comprise one or more of morphological details, texture measures for a marker, intensity measures for a marker, statistical details, and values derived from any of the foregoing of a cell or cell population.

42. The computer program product of claim 40, wherein the instructions for computing the distances in (c) comprise instructions for computing an angle or an inner products of vectors, each from a common center point in the biological feature space, wherein a first vector passes through one of the signatures or points of a first stimulus response path and a second vector passes through one of the signatures or points of a second stimulus response path.

43. The computer program product of claim 40, wherein the instructions for computing the distances in (c) comprise instructions for computing Euclidean distances.

44. A method of determining a separation distance between response paths associated with a plurality of stimuli in a biological feature space, the method comprising:

(a) receiving response paths of two or more stimuli, wherein the response paths represent the responses of one or more biological systems to the stimuli;

(b) aligning the response paths of the stimuli on the basis of distance from a common point in the biological feature space; and

(c) determining a distance between the aligned response paths.

45. The method of claim 44, wherein (c) comprises integrating an expression for distance between aligned points on the aligned response paths.

46. The method of claim 45, wherein the expression for distance is an expression for angle or inner product.