# US 2014/0195165 A1 · Systems and methods for identifying the relationships between a plurality of genes

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0195165 A1
- Document ID: US20140195165A1
- Status: Published application
- Kind: A1
- Application number: 14/078,994
- Filed: 2013-11-13
- Published: 2014-07-10
- Earliest claimed: 2012-11-14
- Assignees: The Translational Genomics Research Institute
- Applicants: The Translational Genomics Research Institute
- Inventors: Seungchan Kim; Sungwon Jung
- Web page: https://patentgrep.com/patent/US20140195165A1

## CPC

- G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
- G16B 5/00: ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
- G16B 5/20: Probabilistic models
- G16B 20/00: ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

The present invention relates to a method and system for the evaluation of differential dependencies of a set of discrete random variables between two conditions. In some embodiments, the system and method compares two conditions by evaluating the probability distributions of the likely dependency networks from random variables.

## Claims

### Claim 1

A computer implemented method of statistical testing for identifying or evaluating relationships between a plurality of genes, the method comprising:
receiving a target gene set containing a plurality of genes across a plurality of conditions;
evaluating each gene as a discrete random variable;
identifying a plurality of likely dependency network structures for the plurality of genes for each condition;
computing a probability distribution of the likely dependency network structures for each condition;
computing overall differential dependency relationships between genes in a target gene set across the plurality of conditions by calculating a difference between each probability distribution of likely dependency network structures among the plurality of conditions; and
identifying a plurality of biological functions and pathways that show genetic relationships across the plurality of conditions using the overall differential dependency relationships.

### Claim 2 (depends on 1)

The method according to claim 1, wherein computing the probability distribution of the likely dependency network structures further comprises computing a posterior probability for each of the dependency network structures.

### Claim 3 (depends on 2)

The method according to claim 2, wherein computing the probability distribution of the likely dependency network structures further comprises computing
Pr

(
g
i
|
D
C
)
=
Pr

(
D
C
|
g
i
)
∑
k
=
1
N

Pr

(
D
C
|
g
k
)
wherein Pr(DC|gi) is a likelihood, g is a dependency network, i is greater than or equal to 1, DC is an observed condition.

### Claim 4 (depends on 2)

The method according to claim 2, wherein computing the posterior probability for each of the likely dependency network structures further comprises using a Bayesian Dirichlet equivalence uniform (BDeu) scoring method.

### Claim 5 (depends on 1)

The method according to claim 1, wherein computing a probability distribution of the likely dependency network structures comprises computing:
Prpropose(i; j|DC)=(1−pij)λ,
for a probable dependency structure, gk, that is proposed for DC, wherein:
λ≧1;
eij and eji are edges between variables Vi and Vj;
a direction of at least one of eij and eji is randomly chosen with a probability of 0.5 when the random choice complies with an acyclic property of a directed acrylic graph (DAG) for gk;
each of a dependency between Vi and Vj is independently evaluated; and
an χ2-test is applied to test the independency between each pair of Vi and Vj to obtain a p-value wherein (pij) (=pji).

### Claim 6 (depends on 1)

The method according to claim 1, wherein computing the overall differential dependency relationships between genes in a target gene set across the plurality of conditions further comprises:
computing a divergence between conditions among the plurality of conditions; and
computing a statistical significance of a condition among the plurality of conditions.

### Claim 7 (depends on 6)

The method according to claim 6, wherein computing the divergence between conditions further comprises measuring the divergence using a Jensen-Shannon (JS) divergence.

### Claim 8 (depends on 6)

The method according to claim 6, wherein computing the statistical significance of a condition further comprises computing the statistical significance with a permutation approach.

### Claim 9

A system comprising: a processor; and a non-transitory computer-readable storage medium storing instructions which, when executed on the processor, perform a method comprising:
receiving a target gene set containing a plurality of genes across a plurality of conditions;
evaluating each gene as a discrete random variable;
identifying a plurality of likely dependency network structures for the plurality of genes for each condition;
computing a probability distribution of the likely dependency network structures for each condition;
computing overall differential dependency relationships between genes in a target gene set across the plurality of conditions by calculating a difference between each probability distribution of the likely dependency network structures among the plurality of conditions; and
identifying a plurality of biological functions and pathways that show genetic relationships across the conditions using the overall differential dependency relationships.

### Claim 10

A method of statistical testing for identifying or evaluating relationships between a plurality of genes, the method comprising:
obtaining a target gene set containing a plurality of genes across a plurality of conditions;
evaluating each gene as a discrete random variable;
identifying a plurality of likely dependency network structures for the plurality of genes for each condition;
determining a probability distribution of the likely dependency network structures for each condition;
determining an overall differential dependency relationships between genes in a target gene set across the plurality of conditions by calculating a difference between each probability distribution of the likely dependency network structures among the plurality of conditions; and
identifying a plurality of biological functions and pathways that show genetic relationships across the plurality of conditions using the overall differential dependency relationships.

### Claim 11 (depends on 10)

The method according to claim 10, wherein determining the probability distribution of the likely dependency network structures further comprises computing a posterior probability for each of the likely dependency network structures.

### Claim 12 (depends on 11)

The method according to claim 11, wherein determining the probability distribution of the likely dependency network structures further comprises computing
Pr

(
g
i
|
D
C
)
=
Pr

(
D
C
|
g
i
)
∑
k
=
1
N

Pr

(
D
C
|
g
k
)
wherein Pr(DC|gi) is a likelihood, g is a dependency network, i is greater than or equal to 1, DC is an observed condition.

### Claim 13 (depends on 11)

The method according to claim 11, wherein determining the posterior probability for each of the likely dependency network structures further comprises using a Bayesian Dirichlet equivalence uniform (BDeu) scoring method.

### Claim 14 (depends on 10)

The method according to claim 10, wherein determining a probability distribution of the likely dependency network structures comprises computing:
Prpropose(i; j|DC)=(1−pij)λ,
for a probable dependency structure, gk, that is proposed for DC, wherein:
λ≧1;
eij and eji are edges between variables Vi and Vj;
a direction of at least one of eij and eji is randomly chosen with a probability of 0.5 when the random choice complies with an acyclic property of a directed acrylic graph (DAG) for gk;
each of a dependency between Vi and Vj is independently evaluated; and
an χ2-test is applied to test the independency between each pair of Vi and Vj to obtain a p-value wherein (pij) (=pji).