Drug discovery methods

• US 10,453,553 B2
• Filed: 06/10/2013
• Issued: 10/22/2019
• Est. Priority Date: 02/04/2002
• Status: Active Grant

• Current Assignee: Qiagen Redwood City, Inc. (Qiagen NV)
• Original Assignee: Qiagen Redwood City, Inc. (Qiagen NV)
• Inventors: Chen, Richard O., Cho, Raymond J., Felciano, Ramon M., Holley, Bret, Patel, Viresh, Richards, Daniel R., Selvarajan, Sushma, Steward, Keith, Schneider, Sara
• Primary Examiner(s): Clow, Lori A.
• Application Number:
  US13/914,419
• Publication Number:
  US 20140121120A1
• Time in Patent Office: 2,325 Days
• Field of Search: None
• CPC Class Codes: G16B 20/00 ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
• CPC Class Codes: G16B 25/00 ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression
• CPC Class Codes: G16B 25/10 Gene or protein expression profiling; Expression-ratio estimation or normalisation
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• CPC Class Codes: G16B 45/00 ICT specially adapted for bioinformatics-related data visualisation, e.g. displaying of maps or networks
• CPC Class Codes: G16B 5/00 ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
• CPC Class Codes: G16B 5/20 Probabilistic models
• CPC Class Codes: G16B 50/00 ICT programming tools or database systems specially adapted for bioinformatics
• CPC Class Codes: G16B 50/10 Ontologies; Annotations

Abstract

Methods for identifying disease-related pathways that can used to identify drug discovery targets, to identify new uses for known drugs, to identify markers for drug response, and related purposes.

Claims

1. A method for evaluating user-supplied genomics data, comprising:

defining a profile model based on a profile definition criterion, wherein a profile in the profile model includes a subset of data contained in a structured database, the structured database comprising findings data, wherein the profile definition criterion comprises any of a combination of genes or gene products that form all or part of a disease related pathway, cells or cellular components, anatomical parts, molecular, cellular, or disease processes and relationships between them, a threshold profile size, findings connectivity metrics, or a combination thereof;

building a library of profiles according to the profile model using the structured database, wherein the findings data of the structured database are structured according to predetermined, causal relationships among gene or gene products, the structured database comprising a knowledge base having a frame-based knowledge representation data model structured according to an ontology having slots and facets that define relationships between different instances in the ontology, and the structured database defining biological relationships that are at least one step removed;

wherein building the library of profiled using the structured database comprises;

querying the structured database to find networks of findings within the structured database that meet the profile definition criterion;

extracting said networks of findings from the structured database;

generating graph data structures based on said networks of findings thereby forming an initial set of profiles;

performing a gene expression sensitivity test on the initial set of profiles;

changing criteria of a subset of the initial set of profiles, based on said gene expression sensitivity test, thereby enlarging one or more profiles in the subset of profiles; and

repeating the steps of performing a gene expression sensitivity test on the enlarged profiles and changing criteria of a subset of the enlarged profiles based on said gene expression sensitivity test, in a recursive manner, based on the profile definition criterion, wherein the repeating results in the library of profiles;

receiving user-supplied genomics data;

scoring each profile of the library of profiles against the user-supplied genomics data using a profile-to-data scoring algorithm;

identifying one or more profiles in the library of profiles that have respective subsets of data that overlap at least a portion of the user-supplied genomics data;

determining, for each such overlapped profile, whether the overlap with the user-supplied genomics data is statistically significant based on a computed P-value, wherein the P-value is a probability measure that indicates the likelihood that the overlap is due to chance; and

presenting a list of one or more overlapped profiles determined to have a statistically significant overlap, wherein the list is ranked according to the P-value computed for each of the one or more statistically significant profiles,

wherein each of the defining, building, receiving, scoring, identifying, determining, and presenting steps are performed by a processing system.

2. The method of claim 1, further comprising:

identifying genomics information concepts related to the user-supplied genomics data; and

pre-generating the library of profiles, wherein the library of profiles is built such that it contains a profile in the structured database for each of a plurality of genomic information having a genomic information type, wherein the profile is generated from said genomic information, with its associated genomic information concept as the profile definition criterion, according to the profile model.

3. The method of claim 2, further comprising:

converting findings from the knowledge base into the structured database, wherein the findings include data, from public or proprietary sources, used to build the knowledge base, and wherein the structured database comprises a graph data structure that associates genomic information concepts, from the structured database, that are related to each other as specified by the findings; and

wherein the querying of the graph data structure to pre-generate the profiles from the library of profiles is based on the profile definition criterion.

4. The method of claim 2, wherein the genomic information type is one of a gene, gene product, and biological process.

5. The method of claim 1, wherein the determining whether the overlap is statistically significant comprises:

computing a probability of the overlap as a function of information contained in the structured database.

6. The method of claim 1, wherein the user-supplied genomics data is differential gene expression data, the method further comprising:

identifying a new use for a known therapy, wherein the gene expression data relates to a biological pathway affected by the known therapy;

prioritizing candidate development compounds for further development, wherein the gene expression data relates to a target of one or more candidate development compounds, and giving higher priority to candidate development compounds on the basis of whether or not they are likely to result in an undesirable effect based on their involvement in other biological pathways as embodied in a profile; and

identifying disease-related pathways, wherein a disease is a side effect of drug therapy, wherein the gene expression data relates to the target affected by the drug therapy and the other biological pathways that are also affected by the drug or the drug discovery target and that result in an undesirable phenotype are embodied in the profile.

7. The method of claim 1, wherein the user-supplied genomics data includes differential gene expression data relating to a disease state, the method further comprising:

validating whether the differential gene expression data are genotypic markers for the disease state according to whether a database-asserted biological association related to the disease state, which is shared among a plurality of overlapped profiles, is statistically significant.

8. The method of claim 1, wherein the profile definition criterion further comprises one or more of a number of genes, organismal process, gene connectivity, edge connectivity, findings source type, experiment contex and tissue consistency criterion.

9. The method of claim 1, wherein in the steps of building the library of profiles, each profile is generated from a seed node.

10. The method of claim 9, wherein the seed node is one of a gene, gene product, biological process, and genomic data type.

11. The method of claim 1, further comprising:

computing a statistical significance for a biological association in the one or more statistically significant profiles.

12. The method of claim 1, wherein the building the library of profiles further comprises:

for each profile generated through the steps of building the library of profiles, selecting a node for the profile based on a number of similar findings in the structured database that link the node to a neighboring node.

13. The method of claim 1, further comprising:

displaying information related to the one or more statistically significant profiles and genomics data using a graphical user interface (GUI).

14. The method of claim 1, further comprising:

annotating the one or more statistically significant profiles with biological associations asserted by the structured database including one or more of a cellular process, molecular process, organismal process, and disease process.

15. The method of claim 14, further comprising:

displaying the biological association using one of a graphical user interface (GUI) and a report.

16. The method of claim 14, wherein the annotating uses classification information found in the ontology.

17. The method of claim 1, wherein the determining whether the overlap of an overlapped profile is statistically significant further comprises:

testing a null hypothesis over a discrete probability distribution, the distribution being a function of the database size, profile sizes, the user-supplied genomics data size, and expression values.

18. The method of claim 1, wherein the building comprises:

generating a plurality of libraries of profiles, wherein each library of profiles corresponds to a different one of a plurality of profile definition criterions.

19. The method of claim 1, wherein the generating the graph data strictures further comprises converting said networks of findings into said graph data structures, wherein the converting comprises converting said network of findings from a literature-based representation, wherein each finding represents a result from a performed experiment, to a biology-based representation, wherein each finding represents a conclusion about biology;

and

wherein the changing the criteria of the profiles of the initial set of profiles further comprises changing the criteria of the profiles which meet a gene expression sensitivity threshold, by adding gene members to the profiles or merging one or more of said profiles.

20. The method of claim 1, wherein the library of profiles detect gene expression changes in terms of gene expression sensitivity of said profiles, wherein the performing of the gene expression sensitivity test on the initial set of profiles comprises testing the profiles for sensitivity in detecting gene expression changes by using a per gene-member sensitivity metric;

and

wherein the repeating the gene expression sensitivity test on the enlarged profiles further comprises repeating the gene expression sensitivity test using said per gene-member sensitivity, metric in a recursive manner until the profiles are enlarged such that the profiles reach a threshold ceiling of number of profile members based on the profile definition criterion.