Methods and systems for identifying a drug mechanism of action using network dysregulation

Abstract

Techniques to identify a mechanism of action of a compound using network dysregulation are disclosed herein. An example method can include selecting at least a first interaction involving at least a first gene, determining a first n-dimensional probability density of gene expression levels for the first gene and one or more genes in a control state, determining a second n-dimensional probability density of gene expression levels for the first gene and one or more genes following treatment using at least one compound, estimating changes between the first probability density and the second probability density, and determining whether the estimated changes are statistically significant.

1 . A system for identifying a mechanism of action of a compound using network dysregulation, comprising:
 one or more processors; and   memory, including computer-executable instructions that, if executed by the one or more processors, cause the system to:   (a) select a tissue-specific regulatory network having a plurality of molecular interactions each connecting a pair of genes;   (b) determine a two-dimensional probability density of gene expression levels for each of the interactions in a control state of the regulatory network;   (c) determine a two-dimensional probability density of gene expression levels for each of the interactions in at least one compound-perturbed state of the regulatory network;   (d) estimate, for each of the interactions, differences between the control probability density and the compound-perturbed probability density; and   (e) determine, where the estimated differences are statistically significant, a set of statistically significant dysregulated interactions, based only on the gene expression data; and   (f) rank the statistically significant dysregulated interactions, to provide a set of candidate mechanism of action (MoA) genes.   
     
     
         2 . The system of  claim 1 , wherein the estimating comprises estimating on an interaction by interaction basis. 
     
     
         3 . system of  claim 1 , wherein the estimating comprises using a Kullback-Leibler divergence. 
     
     
         4 . The system of  claim 3 , wherein determining whether the estimated changes are statistically significant further comprises using a null distribution generated by 10 5  Kullback-Leibler values estimated from random pairs of genes, and providing a P-value for the dysregulation of each edge in the network. 
     
     
         5 - 6 . (canceled) 
     
     
         7 . The system of  claim 1 , wherein the interactions comprise every interaction ending in at least a first gene in the regulatory network. 
     
     
         8 . (canceled) 
     
     
         9 . The system of  claim 7 , comprising determining whether the at least a first gene is dysregulated based at least in part on the significance of each interaction. 
     
     
         10 . The system of  claim 9 , wherein the at least a first gene comprises a plurality of genes, and further comprising determining whether each of the plurality of genes is dysregulated. 
     
     
         11 . (canceled) 
     
     
         12 . The system of  claim 7 , wherein the at least one compound comprises a plurality of compounds, further comprising repeating (a)-(f) for each of the plurality of compound, and identifying two or more compounds with similar pharmacological effect. 
     
     
         13 . A system to identify compounds with similar pharmacological effect, comprising:
 one or more processors; and   memory, including computer-executable instructions that, if executed by the one or more processors, cause the system to:   (a) select a tissue-specific regulatory network having a plurality of molecular interactions each connecting a pair of genes, and having a first interaction involving at least a first gene;   (b) determine a first n-dimensional probability density of gene expression levels for the first gene and one or more genes in a control state of the regulatory network;   (c) determine a second n-dimensional probability density of gene expression levels for the first gene and the one or more genes in a first compound-perturbed state of the regulatory network;   (d) estimate changes between the first probability density and the second probability density;   (e) determine whether the estimated changes are statistically significant;   (f) determine, where the estimated changes are statistically significant, whether the interaction is dysregulated;   (g) repeat (a)-(f) for each of m interactions wherein the m interactions comprise every interaction ending in the first gene in a regulatory network;   (h) determine whether the first gene is dysregulated based at least in part on the significance of each interaction;   (i) repeat (a)-(h) for a plurality of genes;   (j) identify a mechanism of action (MoA) of the first compound by selecting genes that are dysregulated, based only on the gene expression data;   (k) repeat (a)-(j) for a plurality of compound treatments using a plurality of compounds; and   (l) identify two or more compounds with similar pharmacological effect, based on similar MoA.   
     
     
         14 . The system of  claim 13 , wherein the estimating comprising estimating on an interaction by interaction basis. 
     
     
         15 . The system of  claim 13 , wherein the estimating comprises using a Kullback-Leibler divergence. 
     
     
         16 . A non-transitory computer-readable storage medium storing thereon executable instructions that, if executed by one or more processors of a computer system, cause the computer system to at least perform (a)-(f) of  claim 1 . 
     
     
         17 . A non-transitory computer-readable storage medium storing thereon executable instructions that, if executed by one or more processors of a computer system, cause the computer system to at least perform (a)-(l) of  claim 13 .