Scale-free network inference methods

• US 20030144823A1
• Filed: 11/01/2002
• Published: 07/31/2003
• Est. Priority Date: 11/01/2001
• Status: Abandoned Application

• Current Assignee: Gene Network Sciences, Inc
• Original Assignee: Gene Network Sciences, Inc
• Inventors: Periwal, Vipul, Hill, Colin, Fox, Jeffrey J.
• Application Number:
  US10/286,372
• Publication Number:
  US 20030144823A1
• Time in Patent Office: Days
• US Class Current: 703/11
• CPC Class Codes: G01N 2333/4739 Cyclin; Prad 1
• CPC Class Codes: G01N 33/5091 for testing the pathological state of an organism
• 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 5/10 Boolean models
• CPC Class Codes: G16B 5/30 Dynamic-time models

Abstract

Presently disclosed are methods for inferring a network model of the interactions of biological molecules and systems for practicing such methods. The systems and methods described herein provide a systematic and computationally feasible solution to the problem of rational inference of the architecture of biological networks, based on a combination of rational and statistical evaluations of quantitative and qualitative data. These methods constrain the search space of possible networks and, in some embodiments, allow the online addition of new data into an existing model without any interruption in analysis. Additionally, these methods can provide a quantitative evaluation of the confidence levels associated with the putative network architectures discovered thereby. The methods naturally incorporate latent nodes in the search space of possible architectures; hence, the system is also capable of predicting new interactions and/or substrates for the biological systems being studied.

Claims

1. A method of inferring a network model of a process, comprising:

generating a search space of candidate networks;

reducing said search space by eliminating one or more non-fitting candidate networks to form a reduced search space;

testing said candidate networks in said reduced search space against one or more criteria to identify an ensemble of networks; and

modeling said process using said ensemble of networks.

2. The method of claim 1, wherein said reducing further comprises:

identifying one or more scale-free candidate networks in said search space; and

classifying all non-scale-free candidate networks as non-fitting candidate networks based on said identifying.

3. The method of claim 1, wherein said process is a biologic process.

4. The method of claim 1, wherein said ensemble of networks is a Bayesian ensemble.

5. An apparatus for inferring a network model of a process, comprising:

means for generating a search space of candidate networks;

means for reducing said search space by eliminating one or more non-fitting candidate networks to form a reduced search space;

computer means for testing said candidate networks in said reduced search space against one or more criteria to identify an ensemble of networks; and

computer means for modeling said process using said ensemble of networks.

6. The apparatus of claim 5, wherein said reducing means further comprises:

means for identifying one or more scale-free candidate networks in said search space; and

means for classifying all non-scale-free candidate networks as non-fitting candidate networks based on said identifying.

7. The apparatus of claim 5, wherein said process is a biologic process.

8. The apparatus of claim 5, wherein said ensemble of networks is a Bayesian ensemble.

9. A computer system for use in inferring a network model of a process, comprising computer instructions for:

generating a search space of candidate networks;

reducing said search space by eliminating one or more non-fitting candidate networks to form a reduced search space;

testing said candidate networks in said reduced search space against one or more criteria to identify an ensemble of networks; and

modeling said process using said ensemble of networks.

10. The computer system of claim 9, wherein said computer instructions for reducing further comprise computer instructions for:

identifying one or more scale-free candidate networks in said search space; and

classifying all non-scale-free candidate networks as non-fitting candidate networks based on said identifying.

11. The computer system of claim 9, wherein said process is a biologic process.

12. The computer system of claim 9, wherein said ensemble of networks is a Bayesian ensemble.

13. A computer-readable medium storing a computer program executable by a plurality of server computers, the computer program comprising computer instructions for:

generating a search space of candidate networks;

reducing said search space by eliminating one or more non-fitting candidate networks to form a reduced search space;

testing said candidate networks in said reduced search space against one or more criteria to identify an ensemble of networks; and

modeling said process using said ensemble of networks.

14. The computer-readable medium of claim 13, wherein said computer instructions for reducing further comprise computer instructions for:

identifying one or more scale-free candidate networks in said search space; and

classifying all non-scale-free candidate networks as non-fitting candidate networks based on said identifying.

15. The computer-readable medium of claim 13, wherein said process is a biologic process.

16. The computer-readable medium of claim 13, wherein said ensemble of networks is a Bayesian ensemble.

17. A computer data signal embodied in a carrier wave, comprising computer instructions for:

generating a search space of candidate networks;

reducing said search space by eliminating one or more non-fitting candidate networks to form a reduced search space;

testing said candidate networks in said reduced search space against one or more criteria to identify an ensemble of networks; and

modeling said process using said ensemble of networks.

18. The computer data signal of claim 17, wherein said computer instructions for reducing further comprise computer instructions for:

identifying one or more scale-free candidate networks in said search space; and

classifying all non-scale-free candidate networks as non-fitting candidate networks based on said identifying.

19. The computer data signal of claim 17, wherein said process is a biologic process.

20. The computer data signal of claim 17, wherein said ensemble of networks is a Bayesian ensemble.