Methods and systems for clustering biological assay data
AssigneeAgilent Technologies, Inc.
InventorNicholas M. Sampas
Life of this patent
Estimating the term…
Abstract
Methods for clustering a data set from a biological assay are provided. Aspects of the methods include applying statistical analyses to bisected subsets of the data at selected cut-of values to identify one or more break points in the data set. In certain aspects, the statistical analyses employed are based on determining the p-values of two-tailed t-tests calculated using the bisected data at each cut-off value. Aspects of the invention further include computer programming and systems which are configured to cluster data sets from biological assays according to the subject methods.
Claims20
A method of clustering a data set from a biological assay, said method comprising:
- a) obtaining a data set from a biological assay, wherein said data set comprises N measured values;
- b) sorting said data set from low (sample 1) to high (sample N) based on said measured values;
- c) selecting a set of cut-off points, wherein each of said cut-off points (k) is selected from sample 2 to sample N−2 of said data set;
- d) determining, using a computer, the best cut-off point from said set of cut-off points, comprising;
- i) for each cut-off point k, bisecting said data set into a first subset comprising samples 1 to k and a second subset comprising samples (k+1) to N;
- calculating the p-value of the measured values in said first and second subsets using a 2-tailed t-test;
- and calculating a modality score based on said p-value;
- and
- ii) determining the best cut-off point based on said modality score;
- e) clustering said samples into a first and second group at said best cut-off point if said modality score exceeds a threshold value;
- and
- f) providing the result of said clustering step (e) to a user in a user-readable format.
A system for clustering a data set from a biological assay, said system comprising:
- a computer comprising:
- (a) a communication module comprising an input manager for receiving input from a user and an output manager for communicating output to a user;
- (b) a processing module comprising a biological assay data clustering manager configured to:
- (i) receive a biological assay data set comprising N measured values from a user;
- (ii) sort said data set from low (sample 1) to high (sample N) based on said measured values;
- (iii) select a set of cut-off points, wherein each of said cut-off points (k) is selected from sample 2 to sample N−2 of said sorted data set;
- (iv) determine the best cut-off point from said set of cut-off points, comprising;
- 1) for each cut-off point k, bisecting said data set into a first subset comprising samples 1 to k and a second subset comprising samples (k+1) to N;
- calculating the p-value of the measured values in said first and second subsets using a 2-tailed t-test;
- and calculating a modality score based on said p-value;
- and
- 2) determining the best cut-off point based on said modality score;
- v) cluster said samples into a first and second group at said best cut-off point if said modality score exceeds a threshold value;
- and
- vi) provide the result of said clustering step (v) to said user in a user-readable format.
A physical computer program product comprising a computer readable medium carrying one or more sequences of instructions for clustering a data set from a biological assay, 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:
- a) receiving a data set from a biological assay from a user, wherein said data set comprises N measured values;
- b) sorting said data set from low (sample 1) to high (sample N) based on said measured values;
- c) selecting a set of cut-off points, wherein each of said cut-off points (k) is selected from sample 2 to sample N−2 of said data set;
- d) determining the best cut-off point from said set of cut-off points, comprising;
- i) for each cut-off point k, bisecting said data set into a first subset comprising samples 1 to k and a second subset comprising samples (k+1) to N;
- calculating the p-value of the measured values in said first and second subsets using a 2-tailed t-test;
- and calculating a modality score based on said p-value;
- and
- ii) determining the best cut-off point based on said modality score;
- e) clustering said samples into a first and second group at said best cut-off point if said modality score exceeds a threshold value;
- and
- f) providing the result of said clustering step (e) to said user in a user-readable format.
Dependencies are inferred from the claim text.