Autonomous data mining

• US 6,915,282 B1
• Filed: 02/02/2001
• Issued: 07/05/2005
• Est. Priority Date: 10/26/2000
• Status: Active Grant

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Agilent Technologies, Inc.
• Inventors: Conway, Andrew, Eastman, Peter, Snortland, Howard, Eynon, Barrett
• Primary Examiner(s): STARKS, WILBERT L
• Application Number:
  US09/776,244
• Time in Patent Office: 1,614 Days
• Field of Search: 706/12, 706/13, 706/45, 706/48, 706/50, 707/6, 707/1
• US Class Current: 706/12
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

The invention provides a method and system for performing data mining autonomously with regard to a set of data, and formulating hypotheses in response thereto. An autonomous software element (a) collects sets of data, along with collateral data, into a unified extensible database; (b) formulates possibly interesting hypotheses with regard to those data; (c) evaluates hypotheses, thus relating each hypothesis against a probability it could have occurred by chance; (d) rates each hypothesis in response to multiple factors; (e) reports those hypotheses to users, selecting those users who are most likely to be interested and who are most interested in being informed.

Claims

1. A method including

accessing information about a plurality of gene expression values;

generating from said information a hypothesized relationship about genes relative to said information from a relationship between one or more sets of said gene expression values, said hypothesized relationship having a likelihood of being due to chance not exceeding a predetermined value;

retrieving data regarding interests of each agent of a plurality of agents;

identifying at least one agent of the plurality of agents such that the data regarding interests of said at least one agent satisfies at least one predetermined test indicating that said at least one agent has a potential interest in the hypothesized relationship; and

sending information about said hypothesized relationship to said at least one agent.

2. A method as in claim 1, including collecting said information about the plurality of gene expression values from a plurality of external databases.

3. A method as in claim 1, including collecting said information from at least one external database.

4. A method as in claim 1, wherein generating includes

selecting a first set of genes in response to said gene expression values;

selecting a second set of genes in response to data other than said gene expression values;

applying a statistical technique to said first set and said second set; and

confirming said hypothesized relationship in response to applying said statistical technique.

5. A method as in claim 1, wherein generating includes

selecting a set of genes in response to said gene expression values;

comparing the frequencies of the nucleotide sequences upstream from said set of genes to the frequencies of the nucleotide sequences upstream of genes not in said set; and

constructing a hypothesis that is responsive to sequences that have an anomalous frequency distribution.

6. A method as in claim 1, wherein generating includes

examining pathways of genes with respect to gene activation sequences;

extending said pathways using said gene expression information; and

constructing a hypothesized relationship concerning an extension of at least one of said pathways.

7. A method as in claim 1, wherein generating includes

evaluating correlation values with respect to an identified gene or gene sequence so as to determine variations in the behavior of said identified gene or gene sequence;

and

confirming said hypothesized relationship by applying a statistical technique.

8. A method as in claim 1, wherein the step of identifying includes:

rating said hypothesized relationship with a measure of interest by said at least one agent; and

selecting said at least one agent responsive to said measure.

9. A method as in claim 8, wherein the step of selecting includes:

determining a threshold in said measure of interest; and

comparing said threshold and said measure of interest.

10. A method as in claim 8, wherein said measure of interest is periodically reevaluated.

11. A method as in claim 8, wherein said measure of interest is responsive to complexity of said hypothesized relationship, the likelihood of said hypothesized relationship being due to chance, or whether said hypothesized relationship relates to at least one of:

genes one or more researchers have indicated they are interested in, genes for which there are published papers, selected domain-specific knowledge about gene expression.

12. A method as in claim 1, wherein said data regarding interests includes at least one of:

information about interests of said at least one agent, information about published papers, information about researchers, information about relationships between genes.

13. Apparatus including

memory recording information about a plurality of gene expression values;

an autonomous software element disposed on a computing device, said autonomous software element having access to said memory;

said memory recording information about a hypothesized relationship between said gene expression values, said hypothesized relationship having the properties of (1) having a likelihood of being due to chance not exceeding a predetermined value, and (2) satisfying at least one predetermined test indicating that said hypothesized relationship is of potential interest to at least one agent; and

a communication link coupled to said memory and capable of sending information about said hypothesized relationship to said at least one agent.

14. Apparatus as in claim 13, wherein said communication link is capable of collecting said information about the plurality of gene expression values from a plurality of external databases.

15. Apparatus as in claim 13, wherein said communication link is capable of collecting said information from at least one external database.

16. Apparatus as in claim 13, wherein

said hypothesized relationship includes (a) information about a first set of genes, said first set of genes having been selected in response to said gene expression values, and (b) information about a second set of genes, said second set of genes having been selected in response to data other than said gene expression values;

said hypothesized relationship has been confirmed in response to a statistical technique applied to said first set and said second set.

17. An apparatus as in claim 13, including

a means for selecting a first set of genes in response to said gene expression values;

a means for selecting a second set of genes in response to data other than said gene expression values;

a means for applying a statistical technique to said first set and said second set; and

a means for confirming said hypothesized relationship in response to applying said statistical technique.

18. An apparatus as in claim 13, including

a means for selecting a set of genes in response to said gene expression values;

a means for comparing the frequencies of the nucleotide sequences upstream from said set of genes to the frequencies of the nucleotide sequences upstream of genes not in said set; and

a means for constructing a hypothesis that is responsive to sequences that have an anomalous frequency distribution.

19. An apparatus as in claim 13, including

a means for examining pathways of genes with respect to gene activation sequences;

a means for extending said pathways using said gene expression information; and

a means for constructing a hypothesized relationship concerning an extension of at least one of said pathways.

20. An apparatus as in claim 13, including

a means for evaluating correlation values with respect to an identified gene or gene sequence so as to determine variations in the behavior of said identified gene or gene sequence;

and

a means for confirming said hypothesized relationship in response to applying a statistical technique.

21. Apparatus as in claim 13, said memory including

information associating said hypothesized relationship with a measure of interest by said agent;

and

a software comparator coupled to said measure of interest and to a selected threshold.

22. Apparatus as in claim 13, wherein said autonomous software element has access to collateral information other than gene expression values.

23. A memory recording information including instructions, said instructions interpretable by a computing device, said instructions including

an autonomous software element having access to information about a plurality of gene expression values;

a first software element coupled to said information and capable of generating a hypothesized relationship between said gene expression values, said hypothesized relationship having the properties of (1) having a likelihood of being due to chance not exceeding a predetermined value, and (2) satisfying at least one predetermined test indicating that said hypothesized relationship is of potential interest to at least one agent other than said autonomous software element;

a second software element coupled to information about said hypothesized relationship and capable of sending information about said hypothesized relationship to said at least one agent.

24. A memory as in claim 23, including

information about a first set of genes selected in response to said gene expression values;

information about a second set of genes selected in response to data other than said gene expression values;

information about said hypothesized relationship selected in response to applying a statistical technique to said first set and said second set.

25. A memory as in claim 23, wherein generating by the first software element includes

selecting a first set of genes in response to said gene expression values;

selecting a second set of genes in response to data other than said gene expression values;

applying a statistical technique to said first set and said second set; and

confirming said hypothesized relationship in response to applying said statistical technique.

26. A memory as in claim 23, wherein generating by the first software element includes

selecting a set of genes in response to said gene expression values;

comparing the frequencies of the nucleotide sequences upstream from said set of genes to the frequencies of the nucleotide sequences upstream of genes not in said set; and

constructing a hypothesis that is responsive to sequences that have an anomalous frequency distribution.

27. A memory as in claim 23, including

information about examining pathways of genes with respect to gene activation sequences;

information about extending said pathways using said gene expression information; and

information about constructing a hypothesized relationship concerning an extension of said pathways.

28. A memory as in claim 23, wherein the first software element

evaluates correlation values with respect to an identified gene or gene sequence so as to determine variations in the behavior of said identified gene or gene sequence;

and

confirms said hypothesized relationship in response to applying a statistical technique.

29. A method including

accessing information about a set of genes and proteins;

generating from said information a hypothesized relationship involving at least one of the following;

a set of proteins, SNPs and chemicals pertaining to said information, said hypothesized relationship relative to said information and having a likelihood of being due to chance not exceeding a predetermined value;

retrieving data regarding interests of each agent of a plurality of agents;

identifying at least one agent of the plurality of agents such that the data regarding interests of said at least one agent satisfies at least one predetermined test indicating that said at least one agent has a potential interest in the hypothesized relationship; and

sending information about said hypothesized relationship to said at least one agent.

30. A method as in claim 29, including collecting said information from a plurality of external databases.

31. A method as in claim 29, including collecting said information from at least one external database.

32. A method as in claim 29, wherein the step of identifying includes:

rating said hypothesized relationship with a measure of interest by said at least one agent; and

selecting said at least one agent responsive to said measure.

33. A method as in claim 32, wherein the step of selecting includes:

determining a threshold in said measure of interest; and

comparing said threshold and said measure of interest.

34. A method as in claim 33, wherein said measure of interest is periodically reevaluated.

35. A method as in claim 33, wherein said data regarding interests includes at least one of:

information about interests of said at least one agent, information about published papers, information about researchers, information about relationships between genes.