Methods and system for the reconstruction of drug response and disease networks and uses thereof

Abstract

Methods are described that include an integrated, multi-scale, artificial intelligence-based system that reconstructs drug-specific pharmacogenomic networks and their constituent functional subnetworks. The system uses features of the functional topology of the three-dimensional architecture of drug-modulated spatial contacts in chromatin space. Discovery of drug pharmacogenomic networks is performed by selecting candidate SNPs with the aid of imputation, determining predictive causal relationships of the SNPs using machine learning and deep learning, probing spatial genomes as determined by chromosome conformation capture analysis using causal relationship SNPs, combining targeted genes controlled by the same cell and tissue-specific enhancers, and using different data sources and metrics to reconstruct pharmacogenomic networks based on results of genome-wide association studies. The pharmacogenomic networks are deconstructed into their constituent functional and adverse event subnetworks using a knowledge-based segmentation approach for application in clinical decision support, drug repurposing, and in silico drug discovery.

Description

[0001] Cross Reference to Related Applications

[0002] This application claims priority to and the benefit of the following: (1) Provisional U.S. Application Serial No. 62 / 795,705, filed January 23, 2019, entitled “Methods and Systems to Reconstruct Drug pharmacogenomic networks from Pharmacogenomic Regulatory Interactions and Uses Thereof”; and (2) Provisional U.S. Application Serial No. 62 / 795,710, filed January 23, 2019, entitled “Companion Diagnostic Assays for N-methyl-D-Aspartate Receptor Modulators,” the entire disclosure of each of which is hereby expressly incorporated by reference herein. TECHNICAL FIELD

[0003] The technology described herein relates to the discovery of gene network contacts in chromatin space that define the pharmacogenomic substrate of a specific drug, the identification of functionally distinct gene sets within the drug’s pharmacogenomic network (referred to as subnetworks), and the detection of regulatory genomic variants within the drug’s subnetwork gene sets that affect therapeutic efficacy or adverse events. Methods are described that apply these results to characterize drug response in humans for clinical decision support, repurposing of drugs including the development of novel companion therapies, and for in silico drug target discovery. BACKGROUND

[0004] Spatial drug epigenomes, super-enhancer super-enhancers, and topologically associated domains

[0005] New insights into the architecture and dynamics of the non-coding regulatory genome are changing the traditional view of pharmacodynamics and pharmacokinetics. Variations in the non-coding regulatory genome that affect drug response in humans are referred to hereinafter as the "drug epigenome." The drug epigenome can be defined as the active non-coding domains of the human genome that consist of spatial, temporal, and mechanistic regulatory mechanisms of gene regulation in response to xenobiotic stimuli. It contains modulators of gene expression, including enhancers, promoters, and regulatory RNAs, and is characterized by a hierarchy of stereotyped transcriptional domains in which variations greatly affect drug response in humans. Transcriptional control consists of canonical 3D structures that include enhancer-promoter pairs, super-enhancer super-enhancers, transcriptional hubs, mRNA splicing factors, topological associated domains (TADs), and lamina associated domains (LADs). Specific restricted sets of canonical 3D structures are activated or repressed in a cell type specific manner. Drug-disease networks are tightly coupled such that gene variants that are significantly associated with disease are the same, or exist within the same regulatory network that determines drug-based treatment outcomes. Thus, mutations that disrupt the hierarchy of transcriptional space within euchromatin not only convey disease risk, but also concomitant variability in drug response. Pathways containing disease risk, drug response, and concomitant adverse event variants are fertile networks for discovering new drug targets using genotype / phenotype guided computational strategies. These insights better inform patient treatment options based on emerging pharmacological foundations of drug response and drug adverse events. Examples of future therapeutic strategies that involve combination drug design targeting one or more pharmacogenomic networks, integrative multi-scale analysis methods from diverse data type ensembles to enhance stratified drug discovery through molecule-by-molecule environmental modification of the drug epigenome, synthetic editing of non-coding regulatory elements that convey drug treatment resistance, and development of transcription factor-like molecules to cell reprogram tissue damage and atrophy.

[0006] It is important to note that highly significant SNP trait associations from genome-wide association studies (GWAS), phenotype-wide association studies (PheWAS), and other biobank patient data that contain SNPs that convey disease risk, as well as individual responses to specific drugs, all reside in regulatory elements of the non-coding genome known as enhancers. In many cases, enhancers target gene promoters or regulatory RNAs within the same TAD, and can be controlled by larger regulatory elements known as super-enhancers.

[0007] Human genes that play a key role in health, disease states, and drug responses are often regulated by long DNA elements that span 2 or more TADs known as "super-enhancers" or "stretched enhancers" (referred to herein as super-enhancer super-enhancers). Super-enhancer super-enhancers are clusters of enhancers that are occupied by an abnormally high density of interacting factors and activate differential transcription (also known as gene expression) at a higher frequency than typical enhancers. Super-enhancer super-enhancers are multi-molecular assemblies analogous to nucleoli that represent macromolecular condensates that concentrate transcriptional regulation within the nucleus of a cell and segregate it. Super-enhancer super-enhancers occupy known genomic locations that span multiple TADs and LADs in a cell- and development-specific manner.

[0008] Mutations that alter super-enhancer super-enhancers and disrupt their regulation of genes and RNAs, leading to the elimination or alteration of chromatin loops between enhancer-promoter or promoter-promoter pairs, and / or breaking the boundaries of TADs or dispersing the repressor set of TADs known as LADs, have profound effects on the variability of drug responses and the incidence of adverse drug events in the human population.

[0009] The largest pharmacogenomic effect sizes are found in patients with SNPs that are single base changes that disrupt super-enhancers, leading to life-threatening acute adverse drug events. Examples include clozapine-induced agranulocytosis / neutropenia and Stevens-Johnson syndrome or toxic epidermal necrolysis caused by carbamazepine, lamotrigine, phenobarbital, allopurinol, non-steroidal anti-inflammatory drugs, and certain other pharmaceuticals. These adverse drug responses are severe enough that countries such as Singapore and Taiwan require patients to be tested for the presence of these SNPs before administering these pharmaceuticals.

[0010] Super-enhancers are responsible for the identification of different cell types during specified development, and in tissues such as the brain, they serve as a platform for binding of neural-specific transcription factors and mediator complexes. They represent non-traditional pharmacodynamic targets, and their involvement in differential neurogenesis in the adult brain is also a mechanism by which histone deacetylase inhibitors exert their effects in the CNS. Similarly, the unconventional interpretation of drug responses and resolving single nucleotide polymorphisms (SNPs) from GWAS and PheWAS significantly improves the understanding of how mutations in the molecular physiology of cells lead to variations in human pharmacogenomics.

[0011] The spatial hierarchy of transcriptional organization was first determined by chromatin conformation capture methods. Chromatin fills most of the available volume of the nucleus and serves as a chromosome map (CT) and contains restricted A and B compartments composed of euchromatin and heterochromatin, respectively. Typically, compartment A contains euchromatin and more active gene transcription, and compartment B corresponds to heterochromatin and is gene-poor. Compartment B incorporates LADs located at the periphery of the nucleus. These are unique to the chromosome map and appear to be largely invariant features of chromatin organization, as they are not disrupted when using genome editing methods to disrupt TAD or LAD organization. The A and B chromatin compartments of the CT contain approximately 2,450 TADs with an average length of linear sequence of 100 Kbp (kilobases) to 5 Mbp (megabases). TADs were first characterized using chromatin conformation capture methods such as Hi-C, allowing high-resolution studies of enhancer-promoter loops within TADs and TAD boundary proteins.

[0012] Chromatin is contained within the nucleoplasm of differentiated cells as a large, rope-like helix of genomic DNA enclosed in chromatin. CT exists in 3D space, where spatial proximity and chromatin state determine regulatory interactions, not distance as measured in linear DNA sequence. Although CTs overlap to a large extent, there are multiple spatial interactions between different CTs that can be functional. These include complex transcriptional centers composed of multiple genes, regulatory elements including enhancers and promoters, and functionally related DNA-binding proteins such as transcription factors. Trans interactions include inter-chromosomal spatial contacts involving enhancer-promoter interactions or, in some cases, promoter-promoter pairs.

[0013] Drugs alter restricted TADs in a cell type-specific manner

[0014] Most human genomes are divided into approximately 2,450 basic transcriptional units called TADs, but about 5% of expressed genes and functional long-chain non-coding RNAs are not present within these bound 3D structures. TADs are delineated by restricted boundaries, often contain multiple functionally related genes controlled by enhancers within the TAD, and are invariant across all cell types studied to date. With rare exception, enhancers do not cross TAD boundaries unless the TAD boundary is disrupted by a SNP or other genetic variant. Differences in gene expression between different cell types are a function of TADs being activated or repressed in the cell types. TADs exhibit specific histone modifications, are units of DNA replication timing, and specific and restricted sets of TADs and their trans interacting TADs comprise drug responsive and hormone responsive co-regulatory modules. The boundaries of TADs are relatively invariant between different human cell types. The strength of a TAD boundary can be classified into 5 different domains based on the amount of CTCF bound to the boundary and whether super-enhancers are co-located on the TAD boundary.

[0015] Regulatory pharmacogenomics determines pharmacogenomic networks

[0016] From recent studies, some basic principles of pharmacogenomics have emerged: (1) results from GWAS and PheWAS indicate that over 90% of causal single nucleotide polymorphisms (SNPs) are located within regulatory enhancers, while about 5% are located within protein-coding exons; (2) chromatin contact between enhancers and promoters or coordinated promoters always precede both gene transcription of mRNA encoding proteins and alternative splicing in adult human bodies; (3) histone modifications indicate the regulatory state of any given genomic regulatory element or gene; and (4) in all cases studied to date, causal genetic variants exhibit allele-specificity, regardless of whether the cell is diploid, tetraploid, or octaploid. Extensive studies have shown that it is possible to use machine learning algorithms to predict whether a genetic mutation, such as a SNP located within an enhancer, is causal, which have been trained on DNase I hypersensitivity indicating allele-specificity, and other properties of the epigenome including histone modifications associated with enhancers and promoters. The accuracy of the clinical utility of these machine learning applications has been validated using known causal relationship SNPs and comparing them to the output of these software programs.

[0017] Recent studies have shown that there exists a new class of pharmacodynamic and pharmacokinetic master regulator networks in chromatin whose function is to activate and repress interconnected large gene sets that contact in chromatin space. The controllers of these pharmacogenomic regulatory networks represent a new class of druggable targets in the human body that are distinct from the previous representative class of genetic drugs, consisting of writers, readers, and erasers.

[0018] Enhancer and super-enhancer SNPs associated with disease risk and drug response are key to discovering pharmacogenomic networks of drugs

[0019] Causal mutations such as SNPs found within enhancers, promoters, and splice sites greatly alter chromatin state and can be used as "data probes" for discovering pharmacological networks located within the 3D spatial environment of chromatin within the nucleus. While there are published papers describing methods for developing pharmacological networks in complex tissues such as the human brain, the current problems with gene-gene and protein-protein regulatory pathway approaches are that (1) they are based on the assumption that mutations within protein-coding genes and protein-coding exons represent the majority of major biology-related mechanisms, and / or (2) they look for similarity in the structural and catalytic properties of new compounds that mimic FDA-approved drugs for a given indication or match tissue-specific gene expression patterns to those of FDA-approved drugs for a given indication. Recent studies show that neither of these assumptions is very accurate for discovering new psychopharmaceutical candidates that provide better efficacy and fewer side effects than existing drugs. First, the most important SNP trait associations for disease risk and drug response change the function of enhancers located in the non-coding genome rather than proteins, and genetic variants located within introns of genes disrupt intragenic enhancers that can or can not regulate the expression of the genes in which they are located. SNPs located within protein-coding exons typically disrupt alternative splicing of mRNAs or can disrupt enhancers such that any method that a priori predicts that missense SNPs will change protein products is not accurate. Also, the "guilt-by-association" approach used by projects such as the Library of Integrated Network-based Cellular Signatures (LINCS) project is based entirely on gene expression profiles in cell lines as a proxy to discover new drugs for human tissues such as the brain. The complexity of this human tissue requires a more nuanced and comprehensive interrogation method than what is provided by surrogate methods that rely on using cell line-dependent "shot-gun" expression profiles. SUMMARY

[0020] A method and system that uses bioinformatics and computational methods such as machine learning and deep learning to detect regulatory drug networks within the human body. The foundation of these methods is the ability to reveal previously uncharacterized drug pharmacogenomic networks by using mutation interrogations that stratify drug responses in large populations to pharmacogenomic regulatory interactions embedded within the functional three-dimensional (3D) topology of the human genome. These spatial regulatory interactions provide the architecture of the pharmacogenomic networks of most psychotropic and antitumor drugs.

[0021] There now exist knowledge-based methods that can be used to computationally map drug pathways, in lieu of additional experiments in animal and cell models, and without relying on large existing data sets using sophisticated probabilistic inference methods. The knowledge-based methods described herein can be used to reconstruct drug pharmacogenomic networks acting on different cell types and tissues, and deconstruct these networks into components that mediate different on-target and off-target mechanisms of drugs, post-hoc validated using bioinformatics analysis.

[0022] These methods differ from those that require experimental perturbation of the biology of cells or tissues after drug exposure, or those that rely entirely on the centrality of learned machines for pathway mapping. An important part of the process to determine whether a single nucleotide polymorphism (SNP: which can be a single base pair change or short insertion / deletion) is significantly associated with a specific drug response is to use different machine learning algorithms to determine possible mechanistic causality. Nonetheless, the primary mapping method is based on 3D genomic structure and existing knowledge repositories derived from multiple public data sources and / or from experimental data or proprietary data.

[0023] Figure 2 An exemplary model illustrating how the system uses machine learning and deep learning to integrate and process multi-scale data for pharmacogenomic network reconstruction. This strategy for mapping drug networks provides insights into mechanistic on-target and off-target effects, laying the foundation for subsequent preclinical studies;

[0024] Figure 2 A method for detecting drug pharmacogenomic networks within the human body is shown, which can be performed by a server device. The first step of the method comprises extracting important SNPs associated with a specific drug response. Most of these SNPs have been published in genome-wide association studies (GWAS) and phenotype-wide association studies (PheWAS), and there are a large number of unbiased, peer-reviewed scientific publications that can obtain such data. To improve the accuracy of the location of the SNPs, the server device uses the reference to U.S. Patent Application No. 15 / 977,347, filed May 11, 2018, which is incorporated by reference in its entirety. Figure 4BThe automated drug epigenomic informatics pipeline (PIP) described in 4D and 4E, incorporated herein by reference, processes this. Once imputation and annotation have been performed to further characterize the SNPs in the context of their tissue of origin, a variety of accurate and validated machine learning algorithms trained in the causal disease SNP space are applied to determine possible mechanistic causal relationships. In addition, machine learning is used to characterize missense SNPs, synonymous SNPs, and SNPs located within exons that can be splice site donors or acceptors. The output of this pipeline is a set of "permitted" candidate SNP sets that have been shown to stratify drug response to a specific drug of interest within a population. The next method step performed by the server apparatus includes performing spatial genomic classification using these causal SNP sets to locate their target genes within the same TAD as these SNPs in the case of enhancers, and using these enhancer SNPs to locate their top ranking (e.g., top three) statistically significant spatial contacts within the genome by analysis of data sets generated using chromosome conformation capture methods, most typically generated by Hi-C methods. If the causal enhancer SNPs reside in a TAD with an empirically determined strong boundary of strength III-V, characterized by the TAD boundary containing genes involved in drug absorption, distribution, metabolism, and excretion (ADME), then all intronic genes within the TAD controlled by the same enhancer are saved for further evaluation. Similarly, if the top ranking statistically significant contacted "trans TADs" in the spatial genome contain genes controlled by enhancers that are active in the same cell type and / or tissue in which the drug acts, then they are also saved for further evaluation.

[0025] The candidate gene set containing the intronic and trans TAD genes is then evaluated using pathway analysis from, for example, third party software, to determine if the genes in the candidate gene set have known network connections. Those that do form statistically significant interconnections, most typically determined using Fisher's exact test, include the preliminary candidate spatial network gene set expressed for the drug of interest in the tissue of interest. Genes that do not have significant interconnections with other genes are discarded. This includes the preliminary gene set of the spatial network for the specific drug.

[0026] Knowledge-based semi-automated and automated curation is then performed on this set comprising the spatial network for the specific drug to evaluate for added or removed genes. First, each member of the gene set is thoroughly checked in the context of its defined function, including from primary scientific publications that evaluate its function in the context of the specific drug of interest. Second, the linear distance within each gene from its transcription start site and its stop codon is defined as +10 The entire set of known mutations, containing SNPs, variable numbers of tandem repeats, duplications, and large insertions or deletions, are evaluated for their impact on the known mechanism of efficacy and adverse events of the specific drug of interest. Any functional relationship of the physiological processes associated with the efficacy or adverse events of the specific drug of interest are included in the evaluation. This is not limited to the impact of pharmacogenomics on the response to a specific drug. Third, in complex tissues such as the human brain, the expression pattern of each gene is compared to the neuroanatomical substrate where the specific drug of interest is known to act from other studies. For example, in reconstructing the ketamine spatial network, a dataset from 24 functional neuroimaging studies was examined to determine which brain regions are metabolically active following administration of ketamine in humans. Each gene in the preliminary gene set for the ketamine spatial network was examined to see if its expression in the human brain overlaps with the consensus neurograph derived from the 24 functional neuroimaging studies, detailing the neuroanatomical substrate of ketamine action in the human brain. To accomplish this task, the neuroanatomical neurograph for each gene in the human brain was examined from microarray expression and in situ hybridization results from the Allen Brain Science Institute human brain atlas and RNA-seq results from the GTEx project from the National Institutes of Health. Genes whose expression pattern does not fit the consensus neuroanatomical neurograph are discarded.

[0027] The drug pharmacogenomics network gene set is re-evaluated for whether each gene in the gene set has a known network connection using pathway analysis (e.g., by third party software). Genes whose expression in the tissue of interest for the drug of interest forms statistically significant interconnections, most commonly determined using Fisher’s exact test, are included in the final gene set for the specific drug spatial network. Genes that do not have significant interconnections with other genes are discarded. This includes the final gene set for the specific drug spatial network.

[0028] The next step in the method includes applying an iterative gene set optimization tool and an algorithm for organizing the genes of the spatial network into functional subsets of genes, some of which include drug efficacy and drug adverse event subnetworks within the larger gene set. This involves a measure of similarity of input molecules derived from one or more of multiple data sources related to the mechanism of action of the specific drug, converted to standardized human gene nomenclature, and compared to the genes of the drug pharmacogenomics network. The output of this process is the entire gene set for the spatial network of the specific drug, organized into its constituent subnetworks, including efficacy and adverse event subnetworks.

[0029] The next step of the method comprises providing scientific validation of the spatial network of the specific drug organized into its constituent subnetworks using third party software applications such as bioinformatics and biostatistics. These include top ranked (e.g., top five) statistically significant terms from the Gene Ontology or drug product databases such as MedDRA, top ranked (e.g., top five) canonical pathways determined by pathway analysis (such as by commercial or open source pathway analysis software programs), top upstream dysbiotic modulators, and examples of mutational functional impairment of the spatial network and its subnetworks annotated using statistically significant SNP trait associations from GWAS and PheWAS.

[0030] After validation execution, the spatial network of the specific drug and its constituent subnetworks can be stored in a database and provided to a client device for display.

[0031] The spatial network of the drug and its constituent subnetworks can be applied in several contexts. For example, different embodiments are presented in pharmacogenomic decision support for drug selection, drug repurposing, and in silico drug target discovery. One embodiment of clinical decision support is a method of matching a reference drug pharmacogenomic network and its efficacy and adverse event subnetworks selected from the database of such spatial networks to a patient's specific drug efficacy and adverse event subnetworks. This comparison uses methods in deep learning in which a co-training of efficacy measures between the reference subnetworks and patient subnetworks and pattern matching scores is performed. The output is a separate drug efficacy similarity score and drug adverse event subsimilarity score. It should be noted that the trained personnel in the art will recognize that the reference drug pharmacogenomic network and its constituent efficacy and adverse event subnetworks do not represent the optimal profile. Rather, they reflect the full picture of the mechanism of action of the drug, encompassing both the best and worst impact the drug can have on an individual patient.

[0032] An example of drug discovery in silicon is the selection of the gene member PPP1R1B gene in the ketamine spatial network and is controlled by the same enhancer that controls the gene NEUROD2, a gene whose protein product is involved in neurogenesis, and is in significant spatial contact with the trans TAD containing genes DRD2 and ADORA2A. Mapping the gene set interconnected with the PPP1R1B gene and evaluating the gene ontology leading terms and associated canonical pathways using the methods described herein is shown to be significantly involved in pathways of central nervous system (CNS) development, neuronal differentiation, and neurogenesis. In addition, the PPP1R1B gene is expressed in a restricted set of human brain regions including the preputium, nucleus accumbens, and putamen, as is most of the 24 genes that are significantly interconnected with the gene, a neuroanatomical substrate involved in reward and addiction. Finally, PPP1R1B encodes a druggable phosphoprotein defined as a "dual-function signal transduction molecule." Dopaminergic and glutamatergic receptor stimulation modulates its phosphorylation and acts as a kinase or phosphatase inhibitor. As a target of dopamine, this gene can be used as a therapeutic target for neurological and psychiatric disorders. This represents a potential druggable drug target identified using these methods.

[0033] The results of these methods include a spatial network of the drugs ketamine, valproic acid, lithium, lamotrigine, clozapine, and warfarin. Post-hoc validation of these drug pharmacogenomic networks using bioinformatics methods and knowledge-based segmentation of their efficacy and adverse event subnetworks using the methods of the present disclosure are provided. Details about specific efficacy and adverse event subnetworks are also provided to demonstrate the output of the drug pharmacogenomic network identification system. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1A A block diagram of a computer network and system on which an exemplary drug pharmacogenomic network identification system according to the presently described embodiments can operate is shown;

[0035] Figure 1B is a block diagram of an exemplary drug pharmacogenomic network server according to the presently described embodiments that can operate in the system of Figure 1A

[0036] Figure 1C is a block diagram of an exemplary client device according to the presently described embodiments that can operate in the system of Figure 1A

[0037] Figure 2 ​​This paper presents an exemplary model demonstrating how the system uses machine learning and deep learning to integrate and process multi-scale data for pharmacogenomics network reconstruction. This strategy for mapping drug networks provides insights into mechanistic on-target and off-target effects, laying the foundation for subsequent preclinical studies.

[0038] Figure 3 Examples of TADs are shown, which contain gene promoters, enhancers, superenhancers, structuring proteins contained within the TAD boundaries, and subsequent chromatin looping from the promoter to different exons during gene alternative splicing.

[0039] Figure 4A and 4B The drug expansion properties of adjacent TADs, which act by activating enhancers and / or superenhancers to lead to differential gene expression, are demonstrated. Figure 4C This demonstrates that, compared to traditional measures of linkage disequilibrium within a population, the TAD structure of the human genome provides more accurate information about the target gene localization of enhancers and / or superenhancers;

[0040] Figure 5A A "yarn ball" model of human genome chromatin organization within the cell nucleus is shown, including chromatin spatial interactions. Figure 5B A simple drug network is shown in which 3 superenhancers regulate 6 TADs and 4 TADs lack superenhancer regulation, as well as their trans-interactions in the spatial genome after the “yarnball” is exposed to the drug;

[0041] Figure 6 illustrates a simple example of how a SNP located within an enhancer in a network can disrupt the enhancer's contact with one of its target gene promoters in the TAD, thereby leading to adverse drug events in patients within the drug response cohort. Figure 6A This demonstrates how different laboratory methods can be used to obtain measurements from the chromatin spatial interaction set in three dimensions, and how the data can be analyzed in the form of a 2D graph of enhancer-gene promoter interactions. Figure 6B This study describes how the SNP may disrupt the chromatin loop between the enhancer and one of the two gene promoters it regulates within the TAD. This disruption removes the spatial connection between the enhancer and gene promoter 1, leading to gene 1 regulatory dysfunction, which in turn affects the response of this patient and their cohort to the administration of the specific drug of interest.

[0042] Figure 7 It demonstrates the characteristics of spatial genomes, containing several enhancers in each of the TADs located within non-coding genomic DNA (i.e., intergenic or intronic) that selectively activate or inhibit specific function-related genes within the TAD;

[0043] Figure 8A demonstrates the significant association properties of ADME genes with super-enhancers in the human body, and Figure 8B demonstrates that the association between non-coding variants within super-enhancers enhances those that can significantly alter psychotropic drug response;

[0044] Figure 9 demonstrates an example of SNP rs12967143-G (located within an intragenic enhancer in the TCF4 gene) compared to the significance test results of other GWAS SNPs among various neural and non-neural cell types as described using numerical outputs from six different machine learning algorithms used in the analysis (*p < 0.05; **p < 0.01; ANOVA);

[0045] Figure 10 demonstrates that the TAD containing the PK and HLA gene cluster has strong TAD boundaries and is associated with important biological processes as determined by Gene Ontology;

[0046] Figure 11-1 and 11-2 demonstrates a flowchart representing a method for generating a reconstituted drug pharmacogenomics network and corresponding subnetworks of interest containing a human pharmacogenomics SNP input filter, a drug pharmacogenomics network reconstitution engine, and an iterative gene set optimization engine, an output drug efficacy and adverse event subnetwork;

[0047] Figure 12 demonstrates a flowchart representing an exemplary method for iterative gene set optimization to deconstruct a drug pharmacogenomics network into subnetworks;

[0048] Figure 13 demonstrates a flowchart representing an exemplary method for post-hoc validation of a drug pharmacogenomics network and its constituent subnetworks using standardized bioinformatics analysis;

[0049] Figure 14 demonstrates a flowchart representing an exemplary method for error correction of a drug pharmacogenomics network and its constituent efficacy and adverse event subnetworks with individualized patient response data;

[0050] Figure 15-1 and 15-2 demonstrates a flowchart representing an exemplary method for matching a patient's drug efficacy and adverse events with drug efficacy and adverse events of a reference drug pharmacogenomics network using a similarity score for optimizing drug selection in clinical decision support;

[0051] Figure 16AA flow chart is shown representing an exemplary method of drug target identification and drug repositioning for the druggable target PPP1R1B. Figure 16B and 16C A graphical representation of some of the features in the properties of the druggable target PPP1R1B within the neuronal development and antidepressant mechanism subnetwork 2 of the ketamine space network is also shown. Graphical representations of the traits of the ketamine pharmacogenomic network as determined from post-hoc validation of the ketamine pharmacogenomic network and its efficacy and adverse event subnetworks are shown. Figure 16D Gene expression data for key pharmacogenomic efficacy genes in relevant brain tissue regions are shown;

[0052] Figure 17A A general topological model of CNS and peripheral drug response is shown, including chromatin remodeling, PK / hormone regulation, efficacy, adverse events (AEs), systemic PK, and systemic AE and immune system response. Figure 17B Exemplary sets of four pharmacogenomic network topological models defining psychotropic and antineoplastic drug response and their constituent subnetworks are shown, along with these subnetworks and example drugs that fit these topologies. These topologies are used by the systems described herein;

[0053] Figure 18 A graphical depiction of the valproic acid pharmacogenomic network and its constituent subnetworks in the human brain using the methods and systems of the present invention is shown, including chromatin remodeling, efficacy, adverse events, and hormone control and pharmacokinetics;

[0054] Figure 19A The most significant disease annotations for the valproic acid pharmacogenomic network are shown. Figure 19B The top 10 drugs as upstream regulators of the valproic acid pharmacogenomic network are shown. Figure 19C The topological model that fits the valproic acid pharmacogenomic network most accurately is shown;

[0055] Figure 20 An example of the valproic acid pharmacogenomic adverse event subnetwork is shown, and post-hoc bioinformatics analysis indicates that the valproic acid pharmacogenomic adverse event subnetwork is significantly associated with cancer, severe psychological disorders, cognitive impairment, gastrointestinal disorders, lymphoproliferative disorders, movement problems including tremor, and hair loss;

[0056] Figure 21Examples of valproic acid pharmacogenomic neurogenesis subnetworks are shown, and post hoc bioinformatic analysis indicates that valproic acid pharmacogenomic neurogenesis is associated with number of neurons, morphogenesis of neurons, proliferation of neuronal cells, differentiation of neurons, differentiation of embryonic tissues, epilepsy or neurodevelopmental disorders, cognitive impairment, mood disorders, Alzheimer's disease or frontotemporal dementia, and migraine;

[0057] Figure 22 shows examples of disease risk and pharmacogenomic SNPs from GWAS that can be used to determine the predisposition of individual patients to experience adverse events following valproic acid therapy or efficacy response efficacy as shown in Figure 22A Figure 22B

[0058] Figure 23 The output using this system and method is shown overlaid with 4 other experiments and existing data sources including genes that are significantly differentially expressed in the pig (Sus scrofa) brain following peripheral administration of 150 mg / kg of valproic acid and drug databases including Ingenuity Pathway Analysis TM , KEGG, DrugCentral, DrugBank, and LINCS. Note that this system output the largest number of commonly valproic acid-induced genes compared to any of the other 2 comparisons;

[0059] Figure 24 Genes contained within the chromatin remodeling subnetwork of the valproic acid pharmacogenomic network are listed;

[0060] Figures 25-A and 25-B list genes contained within the neuroplasticity and efficacy subnetworks of the valproic acid pharmacogenomic network;

[0061] Figure 26-1 and 26-2 Genes contained within the adverse event subnetwork of the valproic acid pharmacogenomic network are listed;

[0062] Figure 27 Genes contained within the pharmacokinetics and hormones subnetwork of the valproic acid pharmacogenomic network are listed;

[0063] Figure 28A-28I Selected chromatin spatial contacts determined by chromosome conformation capture using the Hi-C method in human neurons are shown for the valproic acid pharmacogenomic network and its functional networks;

[0064] Figure 29A The most significant disease annotations for the ketamine pharmacogenomic network are shown. Figure 29B The top 5 drugs that are upstream regulators of the ketamine pharmacogenomic network are shown.​​Figure 29C A topological model that best fits the ketamine pharmacogenomic network is demonstrated;

[0065] FIG. 30 demonstrates an example gene set enrichment of the output of the gene set optimization engine that distinguished between 2 significantly different subnetworks within the 3 subnetworks comprising the ketamine pharmacogenomic network in the human brain. Figure 30A is the ketamine pharmacogenomic glutamate receptor subnetwork responsible for adverse events associated with the drug and neurotransmission. Figure 30B is the ketamine pharmacogenomic neuroplasticity subnetwork that mediates antidepressant response to ketamine;

[0066] Figure 31 An example of the ketamine pharmacogenomic glutamate receptor subnetwork is demonstrated, and post-hoc bioinformatic analysis indicates that the ketamine pharmacogenomic glutamate receptor subnetwork is significantly associated with the following adverse events (AEs): cognitive impairment, bipolar disorder, postoperative delirium, schizoaffective disorder, schizophrenia, non-cancerous pain, postoperative pain, vomiting, nausea, and unconsciousness;

[0067] Figure 32 An example of the ketamine pharmacogenomic neuroplasticity subnetwork is demonstrated, and post-hoc bioinformatic analysis indicates that the ketamine pharmacogenomic neuroplasticity subnetwork is significantly associated with emotional behavior, morphological abnormality of the nervous system, morphological abnormality of the brain, depression, anxiety, and morphological abnormality of neurons;

[0068] FIG. 33 demonstrates an example of disease risk SNPs from GWAS that can be used to determine an individual patient’s predisposition to experience adverse events following ketamine therapy as Figure 33A indicated or antidepressant response efficacy as Figure 33B indicated;

[0069] Figure 34 A list of genes contained within the neuroplasticity and efficacy subnetwork of the ketamine pharmacogenomic network is listed;

[0070] Figure 35 A list of genes contained within the chromatin remodeling and adverse events subnetwork of the ketamine pharmacogenomic network is listed;

[0071] Figure 36 A list of genes contained within the pharmacokinetics and hormones subnetwork of the ketamine pharmacogenomic network is listed;

[0072] Figure 37A-37G An example of selected chromatin spatial contacts of the entire ketamine pharmacogenomic network determined by chromosome conformation capture using Hi-C in human neurons is demonstrated;

[0073] Figure 38 Neuroanatomical distribution of gene expression data within the ketamine pharmacogenomic network is shown to significantly overlap with localization results from consensus brain maps showing which brain regions are first affected by ketamine;

[0074] Figure 39 Examples of beneficial combinatorial mechanisms and therapeutics discovered using the method of using this system in combination, using valproic acid and ketamine in H3K9 acetylation and deacetylation, respectively, leading to neurogenesis and neural differentiation;

[0075] Figure 40 shows the complementary pharmacogenomic network of valproic acid (Figure 40A) and the pharmacogenomic network of ketamine (Figure 40B) showing neurogenesis and neural differentiation, respectively; Figure 40A Figure 40B

[0076] Figure 41 Show the combined effect of valproic acid and ketamine pharmacogenomic networks in neurogenesis, neuronal proliferation, and terminal neuronal differentiation;

[0077] Figure 42A The most significant disease annotations of the lithium pharmacogenomic network are shown. Figure 42B The top 5 drugs that are upstream regulators of the lithium pharmacogenomic network are shown. Figure 42C The topological model that best fits the lithium pharmacogenomic network is shown;

[0078] Figure 43 The high-resolution partitioning of the gene set subnetworks as an output example using this system for the lithium pharmacogenomic network is shown;

[0079] Figure 44 The genes contained within the chromatin remodeling subnetwork of the lithium pharmacogenomic network are listed;

[0080] Figure 45 The genes contained within the neuroplasticity subnetwork of the lithium pharmacogenomic network are listed;

[0081] Figure 46 The genes contained within the efficacy subnetwork of the lithium pharmacogenomic network are listed;

[0082] Figure 47 The genes contained within the drug-induced weight gain (adverse event) subnetwork of the lithium pharmacogenomic network are listed;

[0083] Figure 48 The genes contained within the drug-induced tremor (adverse event) subnetwork of the lithium pharmacogenomic network are listed;

[0084] ​​Figure 49A The most significant disease annotations for the Lamotrigine Drug Genome network are shown. Figure 49B The top 5 drugs that are upstream regulators of the Lamotrigine Drug Genome network are shown. Figure 49C The topological model that best fits the Lamotrigine Drug Genome network is shown.

[0085] Figure 50 Examples of the Lamotrigine Drug Genome Adverse Event subnetwork as output of this system are shown.

[0086] Figure 51 Examples of the Lamotrigine Drug Genome Neuroplasticity and Efficacy subnetwork as output of this system are shown.

[0087] Figure 52 The genes contained within the Chromatin Remodeling subnetwork of the Lamotrigine Drug Genome network are listed.

[0088] Figure 53-1 and 53-2 The genes contained within the Neuroplasticity subnetwork of the Lamotrigine Drug Genome network are listed.

[0089] Figure 54 The genes contained within the Adverse Event subnetwork of the Lamotrigine Drug Genome network are listed.

[0090] Figure 55 The genes contained within the Pharmacokinetics subnetwork of the Lamotrigine Drug Genome network are listed.

[0091] Figure 56A The most significant disease annotations for the Clozapine Drug Genome network are shown. Figure 56B The top 5 drugs that are upstream regulators of the Clozapine Drug Genome network are shown. Figure 56C The topological model that best fits the Clozapine Drug Genome network is shown.

[0092] Figure 57 Examples of the Clozapine Drug Genome Adverse Event subnetwork as output of this system are shown.

[0093] Figure 58 Examples of the Clozapine Drug Genome Neuroplasticity and Efficacy subnetwork as output of this system are shown.

[0094] Figure 59 The genes contained within the Chromatin Remodeling subnetwork of the Clozapine Drug Genome network are listed.

[0095] Figure 60-1 and 60-2Genes contained within the adverse events subnetwork of the clozapine pharmacogenomic network are listed;

[0096] Figure 61-1 and 61-2 Genes contained within the adverse events subnetwork of the clozapine pharmacogenomic network are listed;

[0097] Figure 62 Genes contained within the pharmacokinetics subnetwork of the clozapine pharmacogenomic network are listed; and

[0098] Figure 63 illustrates the warfarin pharmacogenomic network of any of the network topologies of psychoactive drugs that are not drawn as Figure 11-1 and 11-2 The warfarin pharmacogenomic network is illustrated in Figure 63A and the corresponding gene set enrichment properties are shown in Figure 63B Figure 63C Gene set enrichment of the warfarin anticoagulation subnetwork is shown, and Figure 63D Gene enrichment of the warfarin bleeding and vascular occlusion subnetworks is shown. DETAILED DESCRIPTION

[0099] While the following text sets forth a detailed description of numerous different embodiments, it is understood that the legal scope of the description is defined by the words of the claims set forth in the concluding portion of this patent. The detailed description is to be interpreted in an illustrative manner, and the sequential arrangement of steps in the exemplary process described does not constitute a limitation, but rather is presented as an example of an order in which the described process can occur. Many alternatives, modifications, and variations will be set forth herein.

[0100] It should also be understood that, unless clearly indicated to the contrary, in the patent herein, no part thereof should be construed as being intended to limit the meaning or range of equivalents of any recited term(s) or combination of terms. It is intended that include all structures that would be patentably similar under 35 U.S.C. § 112, sixth paragraph. Lastly, the use of "a" or "an" when used in the context of a patent or claim, means at least one. Accordingly, where "a" or "an" is used preceding the disclosure of a statement or thing, it will be understood that the example total number mores than one, and there can be a plurality of items related to that statement or thing. ​

[0101] This section presents a detailed description of the drug pharmacogenomic network identification system and its application to drug phenotyping decision support for drug selection and pharmacodynamic drug targets within biological pathways. First, a methodological description is presented along with its application in clinical medicine and drug research, followed by several illustrative examples of the drug pharmacogenomic network. The examples are non-limiting and relevant variations of the method will be apparent to those skilled in the art and are intended to be encompassed in the appended claims.

[0102] The drug pharmacogenomic network identification system produces models of the pharmacogenomic regulatory network and its constituent subnetworks using contemporary knowledge bases comprising the functional topology of the pharmacogenomic genomic architecture, the 3D molecular circuitry of control of gene expression and mRNA splicing within chromatin, and the drug-specific geometric expansion and contraction of TADs and their super-enhancer regulation of affected enhancer-promoter and promoter-promoter interactions.

[0103] As shown in Figure 8, the nature of these interactions comprises significant associations of genes encoding proteins involved in the absorption, distribution, metabolism, and excretion (ADME) of xenobiotics drugs - an example consisting of a known mutation in the super-enhancer GH06J032184 responsible for the occurrence of a drug adverse event known as neutropenia in certain individuals after treatment with the antipsychotic drug clozapine.

[0104] The reconstructed drug pharmacogenomic network described herein is inextricably linked to those that mediate disease etiology, thus providing another avenue for studying the mechanisms of pharmacological action. Based on its embedded responsive chromatin plasticity, the drug pharmacogenomic network adapts to intrinsic and extrinsic stimuli over time, which explains pharmacogenomic variability in humans. This determines the response of individual patients to drugs, including drug adverse events, and examples of this variability caused by different proportions of subnetworks within the drug pharmacogenomic network in patients will be provided as examples of output and methods of this system.

[0105] In general, the techniques for identifying a pharmacogenomic network of a drug can be implemented in one or several client devices, one or several network servers, or a system containing a combination of these devices. However, for clarity, the following examples focus primarily on embodiments in which a drug pharmacogenomic network server obtains SNPs from human clinical studies that have been shown to have a significant association with response and adverse events with respect to a particular drug of interest, or it can contain disease or trait risk SNPs. The drug pharmacogenomic network server compares the SNPs to SNPs reported from genome-wide association studies (GWAS), biobanks, phenotype-wide association studies (PheWAS), and other candidate gene studies to identify additional SNPs connected to the SNPs using properties for generating three-dimensional (3D) genomic topology of the set of permissible candidate variants.

[0106] The drug pharmacogenomic network server then performs bioinformatics analysis on each of the permissible candidate variants to filter the set of SNPs into a subset of intermediate candidate variants based on regulatory function, variant dependency, presence of target gene relationships of the permissible candidate variants, and / or whether the permissible candidate variant is a non-synonymous or synonymous coding variant that does not affect the protein but is involved in the regulation of gene expression. In addition, the drug pharmacogenomic network server performs pathway analysis on target genes associated with the subset of intermediate candidate variants to filter the target genes to identify a set of genes causally related to the particular drug of interest.

[0107] The drug pharmacogenomic network server then identifies a pharmacogenomic network of the particular drug of interest based on the identified set of genes and provides an indication of the pharmacogenomic network of the drug for display to a client device. For example, the indication of the pharmacogenomic network of the drug can include a name of the drug of interest, a name and / or graphical depiction of each of the genes in the pharmacogenomic network, and a name and / or graphical depiction of each of the subnetworks within the pharmacogenomic network and the genes in each of the subnetworks.

[0108] The drug pharmacogenomic web server can use various machine learning techniques to analyze the data described herein, such as genomic data and spatial contact data, including but not limited to regression algorithms (e.g., ordinary least squares regression, linear regression, logistic regression, stepwise regression, multivariate adaptive regression splines, locally estimated scatterplot smoothing, etc.), instance-based algorithms (e.g., k-nearest neighbors, learning vector quantization, self-organizing maps, locally weighted learning, etc.), regularization algorithms (e.g., ridge regression, least absolute shrinkage and selection operator, elastic net, least angle regression, etc.), decision tree algorithms (e.g., classification and regression trees, iterative dichotomiser 3, C4.5, C5, chi-squared automatic interaction detection, decision stump, M5, conditional decision trees, etc.), clustering algorithms (e.g., k-means, k-medians, expectation maximization, hierarchical clustering, spectral clustering, mean shift, density-based spatial clustering of applications with noise, order-based points for identifying clustered structure, etc.), association rule learning algorithms (e.g., a priori algorithm, Eclat algorithm, etc.), Bayesian algorithms (e.g., Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Average One-Dependence Estimator, Bayesian Belief Network, Bayesian Network, etc.), artificial neural networks (e.g., perceptron, Hopfield network, radial basis function network, etc.), deep learning algorithms (e.g., multilayer perceptron, deep Boltzmann machine, deep belief network, convolutional neural network, stacked autoencoder, generative adversarial network, etc.), dimensionality reduction algorithms (e.g., principal component analysis, principal component regression, partial least squares regression, Sammon mapping, multidimensional scaling, projection pursuit, linear discriminant analysis, hybrid discriminant analysis, quadratic discriminant analysis, flexible discriminant analysis, factor analysis, independent component analysis, non-negative matrix factorization, t-distributed stochastic neighbor embedding, etc.), ensemble algorithms (e.g., boosting, bootstrap aggregating, AdaBoost, stacked generalization, gradient boosting machine, gradient boosting regression trees, random decision forests, etc.), reinforcement learning (e.g., temporal difference learning, Q-learning, learning automaton, state-action-reward-state-action, etc.), support vector machines, mixture models, evolutionary algorithms, probabilistic graphical models, etc.

[0109] References Figure 1AThe example pharmacogenomic network identification system 100 identifies pharmacogenomic networks for various drugs. The pharmacogenomic network identification system 100 includes a drug pharmacogenomic network server 102 and a plurality of client devices 106-116 that can be communicatively connected over a network 130, as described below. In one embodiment, the drug pharmacogenomic network server 102 and the client devices 106-116 can communicate over wireless signals 120 over a communication network 130, which can be any suitable local or wide area network, including a WiFi network, a Bluetooth network, a cellular network (such as 3G, 4G, Long Term Evolution (LTE), 5G), the Internet, and the like. In some cases, the client devices 106-116 can communicate with the communication network 130 through an intervening wireless or wired device 118, which can be a wireless router, a wireless repeater, a base transceiver station of a mobile phone provider, and the like. By way of example, the client devices 106-116 can include a tablet computer 106, a smart watch 107, a network-enabled cellular phone 108, a wearable computing device (such as Google Glass TM or 109), a personal digital assistant (PDA) 110, a mobile device smart phone 112 (also referred to herein as a "mobile device"), a laptop computer 114, a desktop computer 116, a wearable biosensor, a portable media player (not shown), a phablet, any device configured to communicate wired or wirelessly in RF (radio frequency) communication, and the like. In addition, any other suitable client device that records omics data of a patient, receives pharmacogenomic data sets, or displays an indication of a pharmacogenomic network and / or subnetworks of interest can also communicate with the drug pharmacogenomic network server 102.

[0110] Each of the client devices 106-116 can interact with the drug pharmacogenomic network server 102 to identify a drug of interest for determining a corresponding pharmacogenomic network. Each client device 106-116 can also interact with the drug pharmacogenomic network server 102 to receive an indication of the pharmacogenomic network and / or several pharmacogenomic subnetworks within the pharmacogenomic network of the drug of interest. The client devices 106-116 can present the indication through a user interface for display to a healthcare professional or researcher, such as a display showing a drug-specific topological graph based on the drug-specific topological graph shown in FIG. 17 or the degree of overlap with a similarity score as shown in FIG. 18. Figure 15-1 and 15-2

[0111] ​In example embodiments, the pharmacogenomic web server 102 can be a cloud-based server, an application server, a web server, etc., and includes a memory 150, one or more processors (CPUs) 142, such as microprocessors coupled to the memory 150, a network interface unit 144, and an I / O module 148, which can be a keyboard or a touch screen, for example.

[0112] The pharmacogenomic web server 102 can also be communicatively connected to a database 154 of genomic data, including data from human clinical studies, biobanks, GWAS, and PheWAS studies.

[0113] The memory 150 can be tangible, non-transitory memory, and can include any type of suitable memory modules, including random access memory (RAM), read only memory (ROM), flash memory, other types of persistent memory, etc. The memory 150 can store, for example, instructions for an operating system (OS) 152 that can be executable on the processor 142, which can be any type of suitable operating system, such as a modern smartphone operating system. The memory 150 can also store, for example, instructions for a web reconstruction engine 146A, a pharmacogenomic web bandwidth adjuster 146B, and a gene set optimization engine 146C that can be executable on the processor 142. A more detailed description of the pharmacogenomic web server 102 will be provided below with reference to FIG. 2. Figure 1B A more detailed description of the pharmacogenomic web server 102. In some embodiments, the web reconstruction engine 146A, the pharmacogenomic web bandwidth adjuster 146B, and the gene set optimization engine 146C can be part of one or more of the client devices 106-116, the pharmacogenomic web server 102, or a combination of the pharmacogenomic web server 102 and the client devices 106-116.

[0114] In any case, the network reconstruction engine 146A can receive a request for a drug pharmacogenomic network for a particular drug of interest from the client devices 106-116 or from a database of pre-existing reconstructed drug pharmacogenomic networks. The client devices 106-116 can also provide SNPs from human clinical studies, GWAS studies, PheWAS studies, etc. that have been shown to be significantly associated with response and adverse events with respect to the particular drug of interest or disease risk SNPs from a drug- differentiating pharmacogenomic subnetwork of GWAS. In other embodiments, the network reconstruction engine 146A can obtain the SNPs from the database 154. The network reconstruction engine 146A then generates a method of the permitted candidate variant set based on the obtained SNPs and additional SNPs connected to the obtained SNPs according to TAD boundaries within adjacent TADs modulated by super-enhancers or distant trans interactions determined by chromosome conformation capture methods. Further, the network reconstruction engine 146A performs a bioinformatics analysis on each of the permitted candidate variants to filter the set of SNPs into a subset of intermediate candidate variants and performs a pathway analysis on target genes associated with the filtered set to identify a set of genes causally related to the particular drug. The network reconstruction engine 146A identifies a drug pharmacogenomic network for the particular drug of interest based on the identified set of genes and provides an indication of the drug pharmacogenomic network for display through a user interface of the client devices 106-116.

[0115] The drug pharmacogenomic network server 102 can communicate with the client devices 106-116 over a network 130. The digital network 130 can be a private network, a secure public Internet, a virtual private network, and / or some other type of network, such as a dedicated access line, a plain old telephone line, a satellite link, a combination of these, etc. Where the digital network 130 comprises the Internet, data communications can be conducted over the digital network 130 via Internet communications protocols.

[0116] Turning now to Figure 1BThe pharmacogenomics web server 102 may include a controller 224. The controller 224 may include a program memory 226, a microcontroller or microprocessor (MP) 228, random access memory (RAM) 230, and / or input / output (I / O) circuitry 234, all interconnected via an address / data bus 232. In some embodiments, the controller 224 may also include a database 239, or otherwise communicatively connect to said database or other data storage mechanisms (e.g., one or more hard disk drives, optical storage drives, solid-state storage devices, etc.). The database 239 may contain data such as genomic data, pharmacogenomics web display templates, web page templates, and / or web pages, as well as other data necessary for interaction with users via the network 130. The database 239 may contain data similar to that referenced above. Figure 1A The database described contains 154 similar data.

[0117] It should be understood that, despite Figure 1B Only one microprocessor 228 is depicted, but the controller 224 may contain multiple microprocessors 228. Similarly, the memory of the controller 224 may contain multiple RAMs 230 and / or multiple program memories 226. Although Figure 1B The I / O circuit 234 is described as a single block, but it can contain many different types of I / O circuits. The controller 224 can implement the RAM 230 and / or program memory 226 as, for example, semiconductor memory, magnetically readable memory, and / or optically readable memory.

[0118] like Figure 1B As shown, program memory 226 and / or RAM 230 can store various applications for execution by microprocessor 228. For example, user interface application 236 can provide a user interface to drug pharmacogenomics network server 102, which can, for example, allow a system administrator to configure, troubleshoot, or test various aspects of the server's operation. Server application 238 can operate to receive requests for identifying drug pharmacogenomics networks for a specific drug of interest, identify drug pharmacogenomics networks and pharmacogenomics subnetworks for that specific drug, and transmit instructions for the drug pharmacogenomics networks to client devices 106-116. Server application 238 can be a single module 238 or multiple modules 238A, 238B, such as network reconstruction engine 146A, drug pharmacogenomics network bandwidth adjuster 146B, and gene set optimization engine 146C.

[0119] Despite Figure 1BThe server application 238 is depicted as including two modules 238A and 238B, but the server application 238 can include any number of modules that accomplish the tasks associated with the implementation of the pharmacogenomic web server 102. Further, it should be understood that although only one pharmacogenomic web server 102 is depicted in Figure 1B multiple pharmacogenomic web servers 102 can be provided for distributing server load, serving different web pages, etc. These multiple pharmacogenomic web servers 102 can include web servers, entity-specific servers (e.g., servers for specific pharmaceutical companies, etc.), servers disposed in retail or private networks, etc.

[0120] Referring now to Figure 1C , the laptop computer 114 (or any of the client devices 106-116) can include a display 240, a communication unit 258, user input devices (not shown), and a controller 242, like the pharmacogenomic web server 102. Similar to the controller 224, the controller 242 can include a program memory 246, a microcontroller or microprocessor (MP) 248, a random access memory (RAM) 250, and / or input / output (I / O) circuitry 254, all of which can be interconnected via an address / data bus 252. The program memory 246 can include an operating system 260, a data store 262, a plurality of software applications 264, and / or a plurality of software routines 268. For example, the operating system 260 can include Microsoft Windows®, Linux®, UNIX®, Apple®OS

[0121] OS and / or other data necessary for interacting with the pharmacogenomic web server 102 over the digital network 130. In some embodiments, the controller 242 can also include other data storage mechanisms (e.g., one or more hard disk drives, optical storage drives, solid state storage devices, etc.) resident within the laptop computer 114 or otherwise communicatively connected thereto.

[0122] ​The communication unit 258 can communicate with the pharmacogenomics network server 102 over any suitable wireless communication protocol network, such as a wireless telephony network (e.g., GSM, CDMA, LTE, etc.), a Wi-Fi network (802.11 standard), a WiMAX network, a Bluetooth network, etc. The user input device (not shown) can include a "soft" keyboard displayed on the display 240 of the laptop computer 114, an external hardware keyboard (e.g., a Bluetooth keyboard) that communicates over a wired or wireless connection, an external mouse, a microphone for receiving voice input, or any other suitable user input device. As discussed with reference to the controller 224, it should be understood that although Figure 1C only one microprocessor 248 is depicted, the controller 242 can include multiple microprocessors 248. Similarly, the memory of the controller 242 can include multiple RAMs 250 and / or multiple program memories 246. Although Figure 1C The I / O circuitry 254 is depicted as a single block, but the I / O circuitry 254 can include many different types of I / O circuitry. The controller 242 can implement the RAM 250 and / or the program memory 246 as, for example, semiconductor memories, magnetically readable memories, and / or optically readable memories.

[0123] In addition to other software applications, the one or more processors 248 can be adapted and configured to execute any one or more of a plurality of software applications 264 and / or any one or more of a plurality of software routines 268 that reside in the program memory 246. One of the plurality of applications 264 can be a client application 266 that can be implemented as a series of machine-readable instructions for performing various tasks associated with receiving information at the laptop computer 114, displaying information on the laptop computer, and / or transmitting information from the laptop computer.

[0124] One of the plurality of applications 264 can be a native application and / or a web browser 270 (such as Apple's Google Chrome TM , Microsoft Internet Explorer, and Mozilla Firefox) that can be implemented as a series of machine-readable instructions for receiving, interpreting, and / or displaying web page information from the pharmacogenomics network server 102 while also receiving input from a user such as a healthcare professional or a researcher. Another one of the plurality of applications can include an embedded web browser 276 that can be implemented as a series of machine-readable instructions for receiving, interpreting, and / or displaying web page information from the pharmacogenomics network server 102.

[0125] One of the plurality of routines can include a pharmacogenomic network display routine 272 that obtains an indication of a pharmacogenomic network and presents the indication on the display 240, the indication including a name of a drug of interest, a name and / or graphical depiction of each of the genes in the pharmacogenomic network, and a name and / or graphical depiction of each of the subnetworks within the pharmacogenomic network and the genes within each of the subnetworks. Another of the plurality of routines can include a pharmacogenomic network request routine 274 that obtains a request for a pharmacogenomic network for a particular drug of interest and transmits the request to the pharmacogenomic network server 102.

[0126] Preferably, a user can launch the client application 266 from a client device, such as one of the client devices 106-116, to communicate with the pharmacogenomic network server 102 to implement the pharmacogenomic network identification system 100. Additionally, the user can also launch or instantiate any other suitable user interface application, such as the native application or web browser 270, or any other application of the plurality of software applications 264, to access the pharmacogenomic network server 102 to implement the pharmacogenomic network identification system 100.

[0127] To identify a pharmacogenomic network for a particular drug of interest, the pharmacogenomic network server 102 performs a general method 180 as shown in Figure 11-1 and 11-2 In some embodiments, the method 180 can be implemented in a set of instructions stored on a non-transitory computer-readable memory and executable on one or more processors on the pharmacogenomic network server 102. For example, the method 180 can be performed by the network reconstruction engine 146A, the pharmacogenomic network bandwidth adjuster 146B, and the gene set optimization engine 146C.

[0128] Figure 2 Integration of multi-scale data and processing by the system 100 is shown. The method and system is based on the properties of the pharmacogenomic network composed of enhancer and super-enhancer networks that are activated or repressed in the same human cell type in which the drug first acts. This can be done by matching the changes in gene expression caused by a drug or other therapeutic agent to the placebo or control with higher order structure in which the drug first acts, determining whether SNPs associated with the pharmacogenomic network and subnetworks of the drug are acting in these human tissues, and assessing whether enhancer and super-enhancer regulatory elements are positioned in the restricted target anatomical substrate. Machine learning, deep learning, reinforcement learning, and other methods in artificial intelligence can be used to perform the machine executable steps in the system where applicable.

[0129] A detailed overview of one embodiment of the system 100 is shown in Figure 11-1 and 11-2 This system combines automated executable files with semi-automated curation that includes human pharmacogenomic SNP filters that can accept two SNPs that can influence drug response or disease risk. This is followed by a drug-specific pharmacogenomic network reconstruction engine with drug space network bandwidth adjuster followed by an iterative gene set optimization component that deconstructs the drug-specific pharmacogenomic network into functional subnetworks.

[0130] Selecting genetic variants using human pharmacogenomic SNP input filters

[0131] At block 181, the drug pharmacogenomic network server 102 obtains from human clinical studies SNPs that have been shown to be significantly associated with response and adverse events to a drug of interest, or disease risk SNPs that can be used to distinguish the relative weight of representation, efficacy, and adverse event subnetworks within the regulatory genome of patients. Since the location of SNPs associated with traits under study have been inaccurately assigned to the nearest gene or nearby candidate gene in the literature and reference human genome assembly in most cases, accurate positioning using imputation and annotation techniques is used to determine the actual location of the reported SNPs.

[0132] New studies have several important implications for the drug pharmacogenomic network identification system. First, new drug target mechanisms can be identified by using 3D genomic architecture-based computer vision-based changes in the training set collected using permitted SNPs using deep learning (machine learning) and validation using known drug-induced genome-wide changes in the genomic architecture. Second, clustering new drug target mechanisms in previously defined but not fully informed biological pathways will increase the probability of success. Third, insights gained using 3D genomic architecture from drug response and disease risk SNPs for determining drug targets will lead to next-generation drug candidates, and will greatly improve the accuracy of pharmacogenomic diagnostics.

[0133] Figure 3 The organization of TADs is shown, containing key enhancer and promoter transcription factors CREBBP (CEB-binding protein), EP300 (E1A-binding protein P300), POLR2A (RNA polymerase II subunit A), and YY1 (Yin Yang 1), as well as TAD boundary proteins and chromatin loop-binding proteins Cohesin and CTCF (CCCTC-binding protein). In addition to enhancer-promoter and promoter-promoter pairs, adjacent TADs also contain super-enhancers. The chromatin loop protein CTCF is also involved in the formation of chromatin loops asFigure 3 The splicing of the pre-mRNA is shown.

[0134] Figure 4 illustrates the nature of the drug's effect on TAD regulated by super-enhancers, and TAD-based localization of causal SNP targets provides higher accuracy compared to measures of linkage disequilibrium in the population. Figure 4A and Figure 4B Two adjacent TADs regulated by a super-enhancer are shown. This is in the absence of a drug. Figure 4A In this study, superenhancers were silenced, and differentially expressed genes within adjacent TADs were minimal. Figure 4B It was shown that when the drug is present, it controls two adjacent TADs, causing the TADs to expand geometrically, and the expression of genes located within these expanded TADs is accompanied by increased activation of superenhancers. Figure 4C This demonstrates that TAD organization in spatially regulated genomes provides a more accurate method for locating gene promoter targets of causal enhancer SNPs from GWAS and other studies compared to those offered by traditional measures of linkage disequilibrium.

[0135] Figure 5 shows that the human genome is organized in a 3D manner, similar to a "ball of yarn". This three-dimensional organization changes dynamically over time, but regulatory interactions can be understood by examining the localization of TADs and their regulators (super enhancers) after drug-induced changes.

[0136] In box 182, the pharmacogenomics web server 102 uses a pharmacogenomics informatics pipeline to evaluate candidate causal SNPs. The pharmacogenomics informatics pipeline uses SNPs reported from GWAS, biobanks, PheWAS, and other candidate gene studies to use TAD boundaries instead of Figure 4C The linkage disequilibrium metric shown is used to identify genetically relevant licensed candidate SNPs. The enhancer-regulated SNP workflow evaluates licensed candidate SNPs in disease-related tissues for DNA methylation, transcription factor binding, histone markers, DNase I hypersensitivity, chromatin state, quantitative trait loci (QTLs), chromatin loop-based contacts identified using chromosome conformation capture techniques such as Hi-C, and transcription factor binding site disruption using tissue-specific omics datasets. Figure 11-1 and 11-2 As shown, the drug pharmacogenomics web server 102 then uses an open-source machine learning algorithm to evaluate the final output SNPs to determine whether the SNP is causal or not (box 183), and causal variants are preserved for further analysis in the workflow (box 184). The Altrans algorithm is also used to evaluate exon SNPs as splice donors or splice acceptors. If they are found to be involved in alternative splicing, they are stored accordingly.

[0137] Figure 9 An example of SNP selection for predicted causal relationships using 6 different machine learning and deep learning algorithms based on tissue-specific profiles is shown. This shows a candidate SNP, rs12967143-G, located within an enhancer within the gene for transcription factor 4 (TCF4) gene, with other GWAS SNPs described using the numerical output of the machine learning algorithms used in the analysis. *p < 0.05; **p < 0.01. Numerical scores from each algorithm were generated for each GWAS SNP and only SNPs were retained for further analysis in the case that each output scored the SNP as being a causal relationship in SK-N-SH cells and H1 cells but not HepG2 cells and PBMCs. The scores for each predicted causal relationship SNP were independently tested to determine if the scores were significantly different from scores generated using 10 randomly selected GWAS SNPs using ANOVA at p < 5E-08 listed in the EBI-NHGRI GWAS Catalog for all human traits. Only when a SNP met this criterion of significance was it selected by the system for further analysis.

[0138] Using causal enhancer SNPs to interrogate drug pharmacogenomic networks

[0139] At block 185, enhancer SNPs are used as probes to determine target genes as cis-interactions within the same TAD or adjacent TADs controlled by the same super-enhancer, and drug genomic trans-interactions with other TADs are determined using Hi-C chromosome conformation capture and mapped ChIA-PET datasets generated for the cell types and tissues in which the drug of interest acts (block 187). For drug pharmacogenomic networks, if the TADs involved in cis-interactions and trans-interactions have strong boundaries as predicted by the amount of bound CTCF and / or are significantly associated with a super-enhancer (block 186), genes are selected that contain other functional elements located within the same TAD or adjacent TADs controlled by the same super-enhancer, or in trans-interactions, that are enhancers whose target are significantly changing drug response in the human population, such as long non-coding RNAs. For trans-interactions, if the TADs contain adjacent TADs including the first 3 statistically significant drug genomic contacts of a first set of drug genomic TADs within the same cell and / or tissue type in which the drug of interest acts, and are under the control of the same cell and / or tissue-specific enhancer in which the drug of interest acts (block 188), genes within these “trans-TADs” are selected.

[0140] Figure 7 The distribution of TAD properties in one cell type in the human genome is shown. 98% of TADs contain known or predicted enhancers and 40% of TADs have known super-enhancers that span adjacent TADs in the genome.

[0141] At block 189, the drug pharmacogenomics network server 102 assesses the connectivity of the combined gene set, where the combined gene set is selected from the TAD first set harboring pharmacogenomic SNPs and genes selected from "trans-TADs," including genes that are coherently controlled with the cis-interacting gene first set. For example, the drug pharmacogenomics network server 102 can utilize third party software such as Ingenuity Pathway Analysis TM to examine the connectivity of the combined gene set. Using Fisher's right-tailed exact test, if the drug pharmacogenomics network server 102 determines that there is significant connectivity within the combined gene set based on published literature, the genes are placed into a preliminary gene set for the pharmacogenomic network that includes the drug of interest. Any genes that do not form a connected network are discarded as non-candidate genes for the pharmacogenomic network (block 190).

[0142] Knowledge-based revision using drug pharmacogenomics network bandwidth adjuster

[0143] Manual, semi-automated, or automated curation or a combination thereof is then performed at block 191 on each gene in this genome that comprises the preliminary drug pharmacogenomics network to remove genes whose functions are not relevant to the drug of interest in the cell and / or tissue type in which it acts, or to add other genes that are not part of this preliminary set of the drug pharmacogenomics network that should be added to the set if judged to be affected by the drug of interest in the cell and / or tissue type in which it acts. The interrogation step includes defining the function of the individual gene, the phenotypic consequences of mutational impairment of the gene, and the human cells and tissues in which the gene is expressed to determine whether it can be a candidate for membership in the pharmacogenomic network for the specific drug of interest.

[0144] In one embodiment, these determinations can be made using a combination of manual and semi-automated strategies, incorporating manual curation of each gene, its mutational profile, and its localization of expression within human tissues. These are achieved through various web-based search tools, including gene definitions, genome browser annotations, GWAS catalogs, and other bioinformatics sources. For example, the pharmacogenomics web server 102 can invoke an application programming interface (API) with executable files written in R, Python, PERL, or other programming languages to facilitate data access, data cleaning, and data analysis. This embodiment is an enhanced model of manual curation, but can become time- limited if there are many genes within a gene set or subset of genes of a subnetwork of the pharmacogenomics web, and especially so in cases where functional genomic elements can include regulatory RNAs or functional RNAs such as long non-coding RNAs, or if the function of the gene is difficult to understand. The listing and analysis of the mutational profile of a given gene (+10 Kb upstream and downstream) is the easiest of the 3 interrogation steps to be performed, as these databases are the most comprehensive. There are other sources for the analysis of the tissue distribution of the expression pattern of a gene. In the case where these patterns are compared to the site of action of a specific drug of interest, the pharmacogenomics web identification system 100 can utilize results from imaging modalities, including from radiological studies, optical microscopic analysis in pathology, and even more sophisticated methods. In some embodiments, the pharmacogenomics web server 102 uses machine learning techniques such as neural networks to perform this analysis.

[0145] In another embodiment, the pharmacogenomics web server 102 can use a Bayesian probability classifier based on machine learning or using Bayesian probability calculations. An automated method can be used to reduce the complexity of the data analyzed from different data sources, where the functional knowledge profile of a gene, its mutational profile, and its tissue expression profile are inputs to a learning machine that has been trained on a plurality of such instances and tested independently on another set of instances to determine accuracy. The predictive features selected by the trained neural network can be implemented on a support vector machine classifier to build a functional and mutational prediction model of a gene, where the subsequent machine state determination is with respect to the adequacy of the statistical fit to the pharmacogenomics web.

[0146] In some scenarios, machine learning is prone to overfitting, outputting false positives or false negatives. In another embodiment, the pharmacogenomics web server 102 can use machine learning in parallel to perform semi-automated and Naive Bayes classification to sharpen the accuracy of the final output.

[0147] The drug pharmacogenomics network server 102, and more specifically, the drug pharmacogenomics network bandwidth adjuster 146B, can perform knowledge-based governance using the following steps. First, the drug pharmacogenomics network server 102 examines gene definitions from multiple databases to understand whether they are specifically, rather than generally, affected by the drug of interest. Additionally, when using, for example, Google Scholar... TM Following a thorough internet search using PubMed and / or other sources, published literature containing a text string with the gene name, precursor gene name, or equivalent protein name, along with any function related to the drug of interest, is evaluated. This may include binding affinity studies that reproducibly identify molecules binding with an affinity within 10 times that of the drug of interest to the same pharmacodynamic target. Next, the pharmacogenomics network server 102 examines each gene for all mutations, including SNPs, variable numbers of tandem repeat motifs, duplication, and all other known mutations, extended linearly from the transcription start site and stop codon by 10kb, as examined in a genome browser (such as the UCSC Genome Browser or Ensembl Genome Browser). If any of these mutations are present in published literature or sources such as unpublished clinical trial data, and said mutation is involved in the action of the drug of interest, including efficacy, adverse events, or first-pass metabolism, then said mutation is added to a preliminary gene set including the pharmacogenomics network (box 192). Third, particularly for complex tissues such as the brain, skin, and cardiovascular system, the drug pharmacogenomics network server 102 qualitatively performs consistency mapping to compare the expression of all genes in this final set with the expression (if known) for which the drug of interest exerts its effect. Genes whose expression does not match the pharmacodynamic substrate of the drug of interest are discarded (box 192). Finally, Ingenuity Pathway Analysis, etc., are used. TM Third-party software is then used to examine the connectivity of this gene set (box 193). Using Fisher's right-hand exact test, if the drug pharmacogenomics network server 102 determines that significant interconnectivity exists based on published literature, the gene is placed into the initial gene set of the pharmacogenomics network, which includes the drug of interest. Any gene that does not form a connected network is discarded as a non-candidate gene for the pharmacogenomics network (box 194).

[0148] The drug pharmacogenomics network server 102 can perform knowledge-based governance using gene expression patterns when overlapping indicates functional correspondence. This example is in... Figure 38As shown in the middle, it can exist in the case where the genes of the ketamine pharmacogenomic network exhibit a statistically significant overlap (P < 1E-56; Fisher's exact test) where the drug exerts its rapid effects in the human brain, including the anterior cingulate cortex (ACC) and frontal cortex (FC), but not in the somatosensory cortex (SSC), occipital cortex (OC), or corpus callosum (CC).

[0149] Drug pharmacogenomic network reconstruction engine

[0150] Figure 11-1 and 11-2 The composition of the drug pharmacogenomic network reconstruction engine 146A is shown, which uses proprietary and public knowledge of the 3D human genome, previously defined TADs, super-enhancers, and other properties of the regulatory genome in the original version in the human genome build 19 (hg19) or the newer sparse human genome build 38 (hg38) as provided in the spreadsheet lookup table. This is the key component of the network reconstruction engine 146A. All experimental data generated according to the chromosome conformation capture methods, which can be performed in vivo, are subjected to evaluation, generation of causal SNP probes according to chromatin datasets in cell types of interest where the drug is active, according to the method shown in FIG. 22 and FIG. 33, or from public or other private sources. A parsimonious model of drug-induced changes in the 3D regulatory genome, including TAD matrices, enhancer-promoter pairs, promoter-promoter pairs, and super-enhancers, uses SNPs or candidate variants identified above based on the selected method to develop 2D or 3D forms, as shown in the method in FIG. 6, including 3D modeling of the human genome architecture in Euclidian space, high-resolution light microscopy using FISH, and / or a combination of measures of gene expression (e.g., RNA sequencing, promoter capture Hi-C). After evaluation of the pharmacogenomic interaction set of the drug in chromatin, the resulting pharmacogenomic network is defined initially. To determine whether network elements are significantly interconnected based on existing biomedical knowledge, third-party pathway analysis software is used to provide a significance score. Programs commonly used in gene pathway analysis include Ingenuity Pathway Analysis TM , Panther Gene Ontology pathway mapping, and KEGG (Kyoto Encyclopedia of Genes and Genomes). In this way, the network reconstruction engine 146A determines the interconnections between SNPs and target genes associated with the drug response or adverse events of a particular drug of interest.

[0151] Iterative gene set optimization engine

[0152] At block 195, the drug pharmacogenomic network server 102, and more specifically the gene set optimization engine 146C, performs iterative gene set optimization on the identified candidate gene sets in the pharmacogenomic network for the particular drug of interest. An example method for iterative gene set optimization to deconstruct the drug pharmacogenomic network into subnetworks is illustrated in the flowchart of Figure 12

[0153] Post-hoc validation using third party bioinformatics tools

[0154] ​To scientifically validate the deconstruction of the drug pharmacogenomic network into mechanism subnetworks optimized based on functional gene sets, the drug pharmacogenomic network server 102 performs post-hoc evaluation of each drug pharmacogenomic network's subnetworks for gene ontology leading terms (molecular functions and biological processes), leading terms from drug databases, leading canonical pathways determined, e.g., using other proprietary or open source pathway analysis software, disease risk gene variant analysis determined, e.g., using other proprietary or open source pathway analysis software, and upstream xenobiotic regulators determined using different bioinformatics sources (block 196). The upstream xenobiotic regulators are compared to the particular drug of interest to ensure that the particular drug of interest is the drug most significantly associated with the drug pharmacogenomic network. More specifically, the upstream xenobiotic regulators can be ranked according to their respective association with the drug pharmacogenomic network using different bioinformatics sources. For example, the upstream xenobiotic regulator with the lowest p-value relative to the drug pharmacogenomic network can have the strongest association. The drug pharmacogenomic network server 102 can then determine whether the particular drug of interest is a top ranked upstream xenobiotic regulator or a ranked higher than a threshold rank (e.g., ranked in the top three or ranked in the top five). Additionally, the European Bioinformatics Institute, the National Human Genome Research Institute, and the National Institutes of Health GWAS catalog can be searched to find significant SNP trait associations for each gene in the gene set of each subnetwork. By providing examples of statistically significant SNPs from GWAS, it can provide additional evidence that the mutation burden of the genes included in each subnetwork provides insight into the normal, unimpaired function of the subnetwork. An example method for performing post-hoc validation of the pharmacogenomic network and its constituent subnetworks is illustrated in the flowchart of FIG. 4. Figure 13

[0155] In some embodiments, after performing post-hoc validation, the generated pharmacogenomic network and constituent subnetworks for the particular drug of interest are stored, e.g., in the database 154, as shown in FIG. 5. Figure 1A In some embodiments, the drug pharmacogenomic network server 102 can provide an indication of the pharmacogenomic network and constituent subnetworks to the client devices 106-116 for display to a healthcare professional or researcher. The client devices 106-116 can then present the pharmacogenomic network and constituent subnetworks in the form of a graphical display.

[0156] ​Error correction of drug pharmacogenomic networks and subnetworks

[0157] In some embodiments, the drug pharmacogenomic network server 102 can adjust or tune the pharmacogenomic network and constituent subnetworks for a particular drug to provide accurate models for measuring human drug response phenotypes for real-world clinical applications. From studies of population structure using principal component analysis, allele sharing distance, and other metrics, it has been assumed that the distribution of pharmacogenomic phenotypes can be modeled using a normal distribution, despite the presence of some outliers. For example, cytochrome P450 gene variations that produce differences in CYP450 isoform activity were previously thought to be the main determinants of variability in drug response in humans.

[0158] One embodiment of the present disclosure includes enhancer SNPs within the subnetworks that regulate PK gene expression and other genes located within the same TAD. Variations within these networks can affect tissue-specific metabolism that extends beyond missense codons in the context of drugs and patient dependence. Trans interactions of enhancers that are less constrained by the TADs in which they are located can lead to drug adverse events for patients whose TAD boundaries can be compromised, as exemplified by PK genes.

[0159] Figure 10 Drug metabolism genes and human leukocyte antigen (HLA) genes involved in immune-related drug adverse responses were shown to have the strongest boundaries. Of the 13 gene clusters shown herein, 12 have the strongest TAD boundaries in the human genome (rank V). These include genes encoding cytochrome P450 enzymes (CYP genes), glucuronosyltransferase (UGT) superfamily genes, sulfotransferase (SULT) superfamily genes, N-acetyltransferase (NAT) family genes, and most of the HLA genes. Mutations, such as SNPs, located in the TAD boundaries of these genes have deleterious effects on drug metabolism and drug response variation, including the occurrence of drug adverse events in human populations.

[0160] It is also recognized that additional variables play a role in human drug response, including the effects of social conditions and other environmental factors, which are often difficult to measure.

[0161] FIG. 17 illustrates a pharmacogenomic network topology that can be used by the system as a model of action for most psychoactive drugs. In this example, the network is composed of 5 subnetworks, each of which is associated with a different drug metabolism pathway. The network is shown in a circular topology, with each subnetwork represented by a different color. The network is connected to the drug response phenotype, which is represented by a black circle at the center of the network. The network is connected to the patient population, which is represented by a blue circle at the top of the network. The network is connected to the drug, which is represented by a red circle at the bottom of the network. Figure 17AIn this study, the template model of the central nervous system (CNS) includes drugs and boxes containing: (1) chromatin remodeling; (2) efficacy (EFF) and / or neuroplasticity (NP); (3) CNS adverse events (AEs); and (4) centrally active pharmacokinetic enzymes (PKs) and hormones (Hs). For some drugs, systemic pharmacokinetics (SPKs) are a major determinant of variation in human drug response, and peripheral adverse drug events involving the immune system (IAE) are problematic. Figure 17B The figure shows different pharmacogenomics network topologies of psychotropic drugs with different biospectrums cooperating with different two-dimensional topologies. Therefore, the pharmacogenomics network of drugs can be deconstructed into constituent subnetworks using the subnetwork types shown in Figure 17, where the subnetwork types of most psychotropic drugs may contain two or more of the following: (1) chromatin remodeling; (2) efficacy (EFF) and / or neuroplasticity (NP); (3) CNS adverse events (AE); (4) centrally active pharmacokinetic enzymes (PK) and hormones (H); (5) systemic pharmacokinetics (SPK); and (6) peripheral drug adverse events involving the immune system (IAE).

[0162] Figure 14 A machine learning-based method 600 is demonstrated, which tunes the computationally predicted efficacy and adverse event subnetwork metrics of a drug pharmacogenomics network in a population, including training and test sets, to obtain accurate discretization of response phenotypes. In some embodiments, the drug pharmacogenomics network server 102 performs... Figure 14 Method 600 is illustrated herein. Similarly, in some embodiments, method 600 may be implemented in a set of instructions stored on a non-transitory computer-readable storage medium and executable on one or more processors on the pharmacogenomics network server 102. In any case, method 600 increases the accuracy of the assumed distribution of human response phenotypes for a specific drug developed by computational analysis by training such a drug-specific subnetwork with the pharmacogenomics network identification system 100, wherein the human drug response phenotype is derived from the efficacy and adverse event subnetworks. In this way, the pharmacogenomics network server 102 improves the utility of the distribution of human response phenotypes for use in real-world clinical applications, making it suitable for any reference-based comparative metric performed in medicine or life sciences.

[0163] Matching the pharmacogenomics network of the reference drug with the patient

[0164] The learning architecture used to train the pattern matching subnetwork includes pre-training on a reference set (reference number 710), such as... Figure 15-1 and 15-2 As shown. This reference Figure 15-1and 15-2 The method 700 shown in FIG. 6 can also be further described by the method 700 that can also be performed by the pharmacogenomic web server 102. More specifically, at block 704, the pharmacogenomic web server 102 develops a pattern matching subnetwork of the patient derived from the patient inputted biological sample and co-develops separate trained pattern metrics of the features of the efficacy and adverse event subnetworks to a joint feature representation metric (block 712). To determine similarity to the reference set (blocks 706, 708), the two different reference-patient metric pairs contain an accurate measure of similarity as well as output similarity scores for each of efficacy and adverse events (blocks 714, 716). At block 702, the biological sample obtained from the patient, which can be a cheek swab, blood, or urine sample, undergoes targeted enhancer SNP genotyping as well as combined chromosome conformation capture and RNA-seq. Then at block 704, the pharmacogenomic web server 102 performs the analysis necessary to build the input patient-specific mapping of the efficacy and adverse event subnetworks of the specific drug of interest. These patient-specific, drug-induced subnetwork patterns can be further processed using Bayesian probability calculations to fill in sparse or missing data. When a new patient is inputted as input, the pre-trained reference set for the drug-specific efficacy and adverse event subnetworks for pattern matching is again optimized for the subsequent patient, resulting in a more accurate measure of pharmacogenomic variability between humans with enhanced clinical utility. This matching task assumes that the patches undergo the same feature encoding prior to computing and outputting the similarity scores, greatly increasing efficiency while reducing computational requirements.

[0165] Thus, the feature set extraction and inference using Bayesian distribution based probability calculations on sparse data is constructed differently for each input set (reference set (reference number 710) and patient set (reference number 720)) to increase the accuracy of the reference graph and patient graph. The trained feature network is based on a “Siamese” network approach, with the constraint that the two sets must share the same parameters. When complete, the patient’s drug-induced trained pattern network is coupled with the network obtained from the reference database, paired efficacy feature set pairs, and adverse event feature set pairs. These provide the basis for the development of trained efficacy metrics and trained adverse event metrics that attempt to match the reference set from the patient for all features and the drug of interest. These pairwise matching scores result in separate efficacy and adverse event similarity scores between the reference and patient.

[0166] A further refinement of this embodiment is the development of a reference pattern matching set for each patient that can be used to create a patient-specific database of such reference graphs and updated in a periodic manner as additional biological samples are obtained from the patient in a longitudinal manner, obtained in a clinical setting or outpatient pharmacy over time.

[0167] Methods for identifying drug targets in silicon

[0168] Figure 16A and Figure 16B Another method 800 is shown, which utilizes a pharmacogenomics network of a specific drug of interest, along with efficacy and adverse event subnetworks, to develop molecules as druggable pharmacodynamic targets. In some embodiments, a drug pharmacogenomics network server 102 performs this process. Figure 16A The method 800 shown in the figure. Similarly, in some embodiments, method 800 may be implemented in a set of instructions stored on a non-transitory computer-readable storage medium and executable on one or more processors on the pharmacogenomics web server 102.

[0169] In any case, previously unidentified genes encoding druggable pharmacodynamic targets can be linked to the efficacy subnetwork of a drug-specific pharmacogenomics network, with minimal connectivity to multiple genes in the adverse event subnetwork of the drug-specific pharmacogenomics network at the pharmacogenomics regulatory level. Figure 16A and 16B In the example presented, a ketamine pharmacogenomics network was used, and the druggable target was PPP1R1B (protein phosphatase 1 regulatory inhibitor subunit 1B) (reference number 804)—a bidirectional signal transduction molecule regulated by the neurotransmitter dopamine. The PPP1R1B gene 804 is located within the same TAD as the NEUROD2 (neuronal differentiation 2) gene 802, and both genes are regulated by the same enhancers in both neuronal and astrocyte cell lines. Furthermore, the TAD containing these genes interacts trans-with the TAD containing the DRD2 (dopamine receptor D2) 806 and ADORA2A (adenosine A2a receptor) genes 808, which are also controlled by the same neuronal and astrocyte enhancers in their respective TADs. After using the methods described herein, the PPP1R1B pathway, although not a known pharmacogenomics network, is significantly interconnected in the human brain (p = 1E-88), and seven of these genes are contained in the ketamine pharmacogenomics network, including BDNF, DRD2, GRIA1, GRIN1, GRIN2A, KLF6, and PPP1R1B. Four genes are contained in the ketamine neuroplasticity subnetwork of the ketamine pharmacogenomics network, as shown in Figure 29. Figure 16CThe other four are contained within the glutamate receptor subnetwork of the ketamine pharmacogenomics network. Comparing the location of gene expression in different human brain regions in different genes within the PPPR1B pathway, only 14 are expressed at detectable levels in the human brain. Except for GRIA1, GRIN1, and GRIN2A, which are more broadly expressed in the human brain, the remaining 11 genes in this pathway show a markedly restricted gene expression pattern, limited to the posterior tail, accumbens, and putamen ( Figure 16D ) brain regions.

[0170] At block 810, the drug pharmacogenomics network server 102 performs a bioinformatics analysis of the PPPR1B pathway limited to those 14 genes expressed in the human brain. The bioinformatics analysis shows that the 14 genes are significantly related to neuronal differentiation, neuronal development, and regulation of neurogenesis and CNS development, as well as opioid signaling. These characteristics are common to the neuroplasticity subnetwork of the valproic acid pharmacogenomics network as shown in Figure 18 and 21 the neuroplasticity subnetwork of the ketamine pharmacogenomics network as shown in Figure 32 and the lithium pharmacogenomics network as shown in Figure 43 .

[0171] As shown in Figure 39 , 40, and 41, the drug pharmacogenomics network identification system 100 can reveal complementary properties that can be combined together as new drug compounds, can reveal complementary properties of existing drugs that can be administered sequentially, or can identify similar combinations of drugs from the same drug class to provide a more comprehensive therapy for a given clinical indication. Figure 41 It is shown that the valproic acid neurogenesis pharmacogenomics subnetwork is enriched to stimulate early neurogenesis, and the ketamine neuroplasticity subnetwork is responsible for late neurogenesis. Figure 39 One mechanism by which this combination of therapeutics operates is through sequential acetylation and deacetylation of histone lysine 9 (H3K9) moieties. Thus, valproic acid combines histone deacetylation inhibition with induction of the neural progenitor BAF chromatin remodeling complex for conversion of pluripotent neuronal precursor cells to committed neuronal progenitor cells as shown in Figure 39 (top) and ketamine converts committed neuronal progenitor cells to terminally differentiated neurons by activating HUSH (human silent homologous) and H3K9 methylation Figure 39 (bottom).

[0172] Figure 40 shows gene set enrichment of nuclear genes contained within the valproic acid pharmacogenomics network Figure 40A and the ketamine pharmacogenomics network Figure 40B .Figure 40A It is shown that the valproic acid pharmacogenomic network is enriched for both H3K9 histone deacetylase activity and neurogenesis, and Figure 40B It is shown that the ketamine pharmacogenomic network is enriched for both H3K9 histone methyltransferase activity and neuronal differentiation.

[0173] Thus, as Figure 41 shown, it is possible to combine these approved drugs in clinical use to provide a comprehensive solution for providing early and mid-stage to late neurogenesis, encompassing mechanisms of neurogenesis, neuronal proliferation, neuronal differentiation, and synaptic integration. For example, a first drug (such as valproic acid) can be administered to a patient at a first time point, and then a second drug (such as ketamine) can be administered to the patient at a second time point after the first time point. Thus, this combination of FDA-approved therapeutics can be used not only for disease states in which neuronal cell loss is a characteristic feature of the condition, but also for the aging human brain to maintain the integrity of the gray matter. Disease states can include neurological conditions, neurodegenerative disorders such as frontotemporal dementia, Alzheimer's disease, and Parkinson's disease, and neuropsychiatric conditions including bipolar disorder and schizophrenia, as well as acute brain injury.

[0174] For example, a method for treating a patient having a neurodegenerative disorder can include administering valproic acid to the patient and administering ketamine to the patient. In some embodiments, the method can include obtaining a biological sample of the patient and comparing or having compared the biological sample to one or more SNPs in the valproic acid pharmacogenomic network that are associated with neurogenesis. The method can also include comparing or having compared the biological sample to one or more SNPs in the ketamine pharmacogenomic network that are associated with neuronal differentiation. In response to determining that the patient's biological sample includes an SNP in the valproic acid pharmacogenomic network that is associated with neurogenesis and an SNP in the ketamine pharmacogenomic network that is associated with neuronal differentiation, valproic acid and ketamine can be administered to the patient to treat the patient's neurodegenerative disorder.

[0175] More generally, the pharmacogenomic network identification system 100 can identify the pharmacogenomic network for any number of drugs. The pharmacogenomic network identification system 100 can then compare properties (e.g., drug response phenotypes) associated with genes within the pharmacogenomic network and / or constituent subnetworks of the first drug to properties (e.g., drug response phenotypes) associated with genes within the pharmacogenomic network and / or constituent subnetworks of the second drug to identify complementary properties between the first drug and the second drug. When complementary properties (e.g., early neurogenesis and late neurogenesis) of a drug panel are identified, the drug panel can be repurposed for testing as therapeutic agents against a particular disease or disease state.

[0176] Figure 18 The valproic acid pharmacogenomic network in the human central nervous system and its constituent subnetworks as outputs of the system are shown. The gene subnetworks consist of: (1) chromatin remodeling and H3K9 acetylation; (2) neurogenesis and antiseizure, antimanic, and anti-migraine properties; (3) adverse events; and (4) hormone regulation and pharmacokinetics.

[0177] Figure 19 shows the results of post-hoc bioinformatic analysis of the valproic acid pharmacogenomic network and its accompanying network topology model. Figure 19A The most significant disease annotations of the genes contained in the valproic acid pharmacogenomic network are epilepsy or neurodevelopmental disorders, cognitive impairment, mood disorders, migraine, and mania. It should be noted that valproic acid is indicated for the treatment of simple and complex absence seizures and as an adjunctive treatment for patients with multiple seizure types including absence seizures, treatment of mania in bipolar disorder and other mood disorders, and prevention and reduction of migraine as a monotherapy and adjunctive therapy. The common adverse event of valproic acid is cognitive fog (cognitive impairment).

[0178] Figure 19B The most significant drugs acting as upstream regulators of the valproic acid pharmacogenomic network are shown to include valproic acid (p = 5.20E-114; Fisher's exact test), the HDAC inhibitor trichostatin A (p = 3.21E-35), and nicotine (p = 5.57E-21).

[0179] Figure 19C The valproic acid pharmacogenomic network is shown to fit model network topology label 1, where the deconstructed gene set subnetworks include chromatin remodeling (CR), neuroplasticity and drug efficacy (NP, EFF), adverse events and neurotransmission (AE, NT), and pharmacokinetics and hormone regulation (PK, H). The non-CNS, peripheral system pharmacokinetics (SPK) subnetwork of the valproic acid pharmacogenomic network cannot be determined from the outputs of this system.

[0180] FIG. 28 illustrates an example of trans interactions in 3D chromatin space of valproic acid pharmacogenomics network as output of the methods and systems described herein. Whole genome Hi-C data mapping was performed using SNPs as data probes, including SNPs contained within the valproic acid subnetwork and obtained from GWAS catalogs, including those significantly associated with disease risk as well as valproic acid response variants and dissociation. These results were used to detect both cis and trans interactions with other members of the valproic acid pharmacogenomics pathway within human neurons. Figure 28A A whole genome map is shown that is useful for understanding Figure 28B-28I Key gene-gene interactions as determined by the Hi-C method are shown in Figure 28B GABBR1 (a gene encoding a receptor for gamma-aminobutyric acid (GABA), the main inhibitory neurotransmitter in the human CNS, and enhancer mutations in this gene are the basis for brain disorders such as epilepsy) is shown in Hi-C contact with CRHR1 and CRHR1-IT1 (genes important for neuroprogenitor differentiation under the control of several super-enhancers, corticotropin hormone binding in the adult brain, and mutations associated with anxiety and manic disorders). Figure 28C GABRG2 (a gene encoding a GABA receptor whose mutations are significantly associated with febrile and infantile seizures) is shown in Hi-C contact with KCNJ3 (a gene encoding a potassium channel in the human CNS whose mutations are significantly associated with cognition and epilepsy) in human neurons space. Figure 28C RUNX1, a transcription factor, is shown in spatial Hi-C contact with HDAC9, a member of the histone deacetylase superfamily, in human neurons. Mutations in the enhancers and super-enhancers of both genes are significantly associated with alopecia and light hair loss, an adverse event associated with valproic acid therapy. Figure 28E GABRB2, GABRG2 (genes encoding GABA receptors and whose mutations are associated with ataxia and epilepsy) are shown in spatial Hi-C contact with KCNQ5 (a gene whose mutation in a super-enhancer is significantly associated with autosomal developmental retardation and intellectual disability) in human neurons. Figure 28F NEUROD1 (a master transcription factor involved in neurogenesis and implicated in valproic acid response in the human body) is shown in spatial Hi-C contact with NEUROG3 (a transcription factor involved in neuroprogenitor commitment to neuronal cell lineage) in human neurons. Figure 28GSpatial Hi-C contacts between GRIN2A (gene encoding N-methyl-D-aspartate (NMDA) receptor member, mutations of which are significantly associated with schizophrenia, bipolar disorder, and mania) and ANK3 (gene encoding synaptic cytoskeletal member and mutations of which are significantly associated with bipolar disorder, sleep type, and schizophrenia) in human neurons are shown. Figure 28H Spatial Hi-C contacts between GRIN2B (gene encoding NMDA receptor member) and SNCA (synaptic presynaptic protein and gene whose super-enhancer mutations are significantly associated with late-onset Parkinson’s disease) in human neurons are shown. Figure 28I Spatial Hi-C contacts between PAX6 (gene encoding master transcription factor responsible for early development of the human central nervous system, eyes, and nose) and SOX2 (gene related to PAX6 that controls the neural progenitor BAF remodelling complex responsible for neurogenesis) in human neurons are shown.

[0181] Figure 20 A valproic acid pharmacogenomics adverse event subnetwork is demonstrated, and post hoc bioinformatics analysis shows that the valproic acid adverse event subnetwork is significantly associated with cancer, severe psychiatric disorders, gastrointestinal disorders, tremor, and alopecia.

[0182] Figure 21 A valproic acid pharmacogenomics neuroplasticity subnetwork is demonstrated, and post hoc bioinformatics analysis shows that the glutamate receptor subnetwork is significantly associated with neurogenesis, neuronal differentiation and proliferation, and disease states including epilepsy, mood disorders, and migraine.

[0183] FIG. 22 demonstrates an example of 237 unique GWAS disease risk and pharmacogenomic SNPs from the efficacy and adverse event profile that can be used to distinguish individual patients’ response to valproic acid’s antiepileptic, antimanic, and anti-migraine properties. Figure 22A GWAS disease risk SNPs located within the valproic acid adverse event subnetwork are shown that can be annotated to enhancers and super-enhancers associated with alopecia, gastrointestinal disorders mediated by the CNS, and reduced efficacy of the mixed antidepressant bupropion in bipolar depression. Figure 22B The valproic acid pharmacogenomics neuroplasticity subnetwork is shown to contain multiple GWAS disease risk and pharmacogenomic SNPs that can be annotated to enhancers and super-enhancers that are significantly associated with chronic migraine, epilepsy, bipolar disorder (International Classification of Diseases (ICD) codes F31.0-F31.64), and valproic acid’s efficacy in bipolar mania.

[0184] Figure 23Validation of the system is shown by comparison to experimental results and public data contained in the most widely used open source and commercial drug databases. Venn diagrams are outputs of the valproic acid pharmacogenomic network genes in human neural tissue compared to differentially regulated genes by valproic acid after administration of the drug to control pig (wild boar) brains, and Ingenuity Pathway Analysis (IPA; Qiagen; GmBH) and Kyoto Encyclopedia of Genes and Genomes (KEGG) and DrugBank (LINCS database from the National Institutes of Health) and DrugCentral regulated gene sets. The valproic acid pharmacogenomic network gene sets are outputs of the system described herein exhibit the highest degree of overlap in all one-to-one comparisons of results and data from all of these sources. TM (IPA; Qiagen; GmBH) and Kyoto Encyclopedia of Genes and Genomes (KEGG) and DrugBank (LINCS database from the National Institutes of Health) and DrugCentral regulated gene sets. The valproic acid pharmacogenomic network gene sets are outputs of the system described herein exhibit the highest degree of overlap in all one-to-one comparisons of results and data from all of these sources.

[0185] Figure 24 is a list of all genes contained in the chromatin remodeling (CR) subnetwork of the valproic acid pharmacogenomic network in the human brain.

[0186] Figures 25-A and 25-B are a list of all genes contained in the neural plasticity and drug efficacy (NP, EFF) subnetwork of the valproic acid pharmacogenomic network in the human brain.

[0187] Figure 26-1 and 26-2 is a list of all genes contained in the adverse events and neurotransmission (AE, NT) subnetwork of the valproic acid pharmacogenomic network in the human brain.

[0188] Figure 27 is a list of all genes contained in the pharmacokinetics and hormone regulation (PK, H) subnetwork of the valproic acid pharmacogenomic network in the human brain.

[0189] Figure 29 shows the results of post-hoc bioinformatic analysis of the ketamine pharmacogenomic network and its accompanying network topological model. Figure 29A The most significant disease annotations for genes contained in the ketamine pharmacogenomic network are shown to be schizophrenia, treatment-resistant depression, bipolar disorder, postoperative delirium, and postoperative pain.

[0190] Figure 29B The most significant drugs acting as upstream regulators of the ketamine pharmacogenomic network are shown to be ketamine (p-value = 6.26E-33 ketamine; Fisher’s exact test), morphine (p = 1.97E-17), and nicotine (p = 6.62E-17).

[0191] Figure 29C A topological tag 2 of the ketamine pharmacogenomics network wiring model is shown, where the resolved gene set subnetworks include chromatin remodeling (CR), adverse events (AE) and neurotransmission (NT), neuroplasticity and drug efficacy (NP, EFF), and pharmacokinetics and hormone regulation (PK, H). The non-CNS, peripheral system pharmacokinetics (SPK) subnetwork of the ketamine pharmacogenomics network cannot be determined from the output of this system.

[0192] Figure 37 illustrates examples of trans interactions of the ketamine pharmacogenomics network in 3D chromatin space as output of the methods and systems described herein. Whole genome Hi-C data mapping was performed using SNPs as data probes, including SNPs contained within the ketamine subnetwork and obtained from GWAS catalogs, including those significantly associated with disease risk and ketamine antidepressant response variants and dissociation. These results validate the pathway analysis and show both cis and trans interactions with other members of the ketamine pharmacogenomics pathway within human neurons. These pharmacogenomics contacts are significantly enriched for association with specific super-enhancers from the cingulate and frontal cortex. Figure 37A A whole genome map is shown that is key to understanding Figure 37B-37G The gene-gene interactions shown in Figure 37B Hi-C contacts between RASGRF2, a gene associated with synaptic plasticity and alcoholism, and the nicotinic receptor genes CHRNA3 and CHRNA5, which co-localize with SNPs in GWASs that are significantly associated with smoking status. Figure 37C Trans interactions between the ROBO2 gene, which contains multiple SNPs associated with unipolar depression in GWASs and both dissociation and antidepressant response to ketamine, and both the GRIN2B gene and the ATF7IP gene. The ATF7IP gene encodes a chromatin remodeling protein essential for methylation of histone 3 lysine 9 (H3K9me3) responsible for heterochromatin formation and gene silencing mediated by HUSH as part of the stable SETDB1 complex. Figure 37D Pharmacogenomic contacts between TCF4 and the GRM5 gene, which encodes a member of the metabotropic glutamate receptor family and contains an enhancer significantly associated with depression in GWASs, are shown. Figure 37E How the Hi-C plot of interactions between CACNA1C and the GRIN2A and ATF7IP2 genes. In Figure 37FHi-C pharmacogenomic contacts obtained from human glutamatergic neurons show trans interactions between the CAMK2A gene located on chromosome 5 and the genes GRIN1 and ANAPC2 located on chromosome 9. The ANAPC2 protein is part of a complex that controls the formation of synaptic vesicles clustered at the active zone into presynaptic membranes in post-mitotic neurons, and this complex also degrades NEUROD2 as a major component of presynaptic differentiation during neuronal differentiation. Figure 37G Pharmacogenomic contacts between the DRD2 gene in neurons and the RHOA gene encoding a signaling protein that regulates cytoskeleton in neurons during synaptic transmission are shown.

[0193] Figure 38 Overlap between genes of the ketamine pharmacogenomic network in the regions of the human brain where ketamine first exerts its rapid antidepressant response and post-mortem human brain is shown. This provides additional evidence supporting the ketamine pharmacogenomic network and can be demonstrated by comparing the neuroanatomical distribution of gene expression data within the ketamine subnetwork with the localization results of a consensus brain map showing which brain regions are first affected by ketamine from 24 neuroimaging studies. The consensus map highlights the anterior cingulate cortex (ACC), dorsal and ventral medial prefrontal cortex (PFC), and supplementary motor area (SMA) as the first brain regions consistently activated by the drug. However, other CNS regions in neuroimaging studies are reported to be rapidly affected by ketamine in humans after drug administration. These in Figure 38 brain regions that are not included in the neuroimaging studies examined during the study period that reported the vast majority of brain regions first affected by ketamine. As a control, adjacent human brain regions not affected by ketamine were selected in neuroimaging studies, including the corpus callosum (CC), occipital cortex (OC), and somatosensory cortex (SS). As shown in Figure 38 ketamine network are expressed at significantly higher levels in the ACC and PFC compared to the adjacent CC, SS, and OC where there is no evidence that ketamine exerts a rapid antidepressant effect. The ACC is part of the cingulate cortex and the PFC is part of the frontal cortex.

[0194] Figure 30 demonstrates the output of the iterative gene set optimization analysis of the ketamine pharmacogenomic network in the human brain. The larger ketamine pharmacogenomic network contains 3 subnetworks, resulting in 2 distinct subnetworks. The glutamate receptor subnetwork is enriched for synaptic signaling, glutamate receptor signaling, glutamate pathway regulation, and chromatin organization. The top out-of-class substance (chemical drug) upregulator of the glutamate receptor subnetwork is ketamine (p = 2.1E-09). The top out-of-class substance (chemical drug) upregulator of the dopamine receptor subnetwork is cocaine (p = 1.1E-06). The top out-of-class substance (chemical drug) upregulator of the dopamine receptor subnetwork is cocaine (p = 1.1E-06). Figure 30A). In contrast, the neuroplasticity subnetwork was enriched for regulation of nervous system development, regulation of neurogenesis, regulation of neuronal differentiation, neurogenesis and nervous system development Figure 30B The neuroplasticity subnetwork exhibited significant enrichment for glutamate receptors as determined by Ingenuity Pathway Analysis TM Significant overlap with the "Cardiovascular Disease, Neurological Disease and Body Injury Abnormality" network class was determined at p = 1E-59, and its top-ranking xenobiotic upregulator was also ketamine at p = 6E-12.

[0195] Figure 31 The ketamine pharmacogenomics glutamate receptor subnetwork was exhibited, and post-hoc bioinformatics analysis showed that the glutamate receptor subnetwork was significantly associated with cognitive impairment, bipolar disorder, postoperative delirium, schizoaffective disorder, schizophrenia, non-cancerous pain, postoperative pain, emesis, nausea, and unconsciousness.

[0196] Figure 32 The ketamine pharmacogenomics neuroplasticity subnetwork was exhibited, and post-hoc bioinformatics analysis showed that the neuroplasticity subnetwork was significantly associated with emotional behavior, morphological abnormality of nervous system, morphological abnormality of brain, depression, anxiety, and morphological abnormality of neuron.

[0197] Figure 33 exhibits an example of 108 unique GWAS disease risk and pharmacogenomics SNPs that can be used to differentiate individual patient response to efficacy and adverse event profiles of antidepressants as well as ketamine and other effects of ketamine analogs. Figure 33A It is shown that multiple GWAS disease risk SNPs located within the glutamate receptor subnetwork can be annotated as enhancers associated with tobacco smoking status, chronic schizophrenia (ICD diagnostic code F20), and bipolar 1 disorder (ICD diagnostic code F31.0 - F31.64). Figure 33B It is shown that the ketamine neuroplasticity subnetwork contains multiple GWAS disease risk SNPs that can be annotated as enhancers associated with recurrent depression (ICD code F33), alcoholism, and response to ketamine.

[0198] Figure 34 is a list of all genes contained in the neuroplasticity and drug efficacy (NP, EFF) subnetwork of the ketamine pharmacogenomics network in the human brain.

[0199] Figure 35 is a list of all genes contained in the chromatin remodeling (CR), adverse events (AE), and neurotransmission (NT) subnetwork of the ketamine pharmacogenomics network in the human brain.

[0200] Figure 36is a list of all genes contained in the pharmacokinetic and hormonal regulation (PK, H) subnetwork of the ketamine pharmacogenomic network in the human brain.

[0201] Figure 42 shows the results of post-hoc bioinformatic analysis of the lithium pharmacogenomic network and its accompanying network topological model. Figure 42A The most significant disease annotations for genes contained in the lithium pharmacogenomic network are shown to be cognitive impairment, mood disorder, drug-induced tremor, drug-induced weight gain, and schizophrenia.

[0202] Figure 42B The most significant drugs acting as upstream regulators of the lithium pharmacogenomic network are shown to include lithium chloride (p-value = 2.23E-23; Fisher's exact test), lithium (p = 4.20E-19), and fluoxetine (p = 2.94E-14).

[0203] Figure 42C The lithium pharmacogenomic network is shown to fit model network topological label 4, where the deconstructed gene set subnetworks include chromatin remodeling (CR), neural plasticity (NP), efficacy (EFF), adverse event (AE), and adverse event (AE). The non-CNS, peripheral system pharmacokinetic (SPK) subnetwork of the lithium pharmacogenomic network cannot be determined from the output of this system.

[0204] Figure 43 A high resolution partitioning of the gene set subnetworks is shown as an example of the detailed output using this system for the lithium pharmacogenomic network.

[0205] Figure 44 is a list of all genes contained in the chromatin remodeling (CR) subnetwork of the lithium pharmacogenomic network in the human brain.

[0206] Figure 45 is a list of all genes contained in the neural plasticity (NP) subnetwork of the lithium pharmacogenomic network in the human brain.

[0207] Figure 46 is a list of all genes contained in the efficacy subnetwork (EFF) of the lithium pharmacogenomic network in the human brain.

[0208] Figure 47 is a list of all genes contained in the drug-induced tremor, adverse event (AE) subnetwork of the lithium pharmacogenomic network in the human brain.

[0209] Figure 48 is a list of all genes contained in the drug-induced weight gain, adverse event (AE) subnetwork of the lithium pharmacogenomic network in the human brain.

[0210] Figure 49 shows the results of post-hoc bioinformatic analysis of the lamotrigine pharmacogenomic network and its accompanying network topological model. Figure 49A The most significant disease annotations for genes contained within the lamotrigine pharmacogenomic network were shown to be epilepsy, fibromyalgia, bipolar 1 disorder, mania, and treatment-resistant schizophrenia.

[0211] Figure 49B The most significant drugs acting as upstream regulators of the lamotrigine pharmacogenomic network were shown to include lamotrigine (p = 1.08E-10; Fisher's exact test), carbamazepine (p = 3.51E-08), and mirtazapine (p = 1.06E-07).

[0212] Figure 49C The lamotrigine pharmacogenomic network was shown to fit model network topological label 2, where the disassembled gene set subnetworks included chromatin remodeling (CR), adverse events (AEs), and neurotransmission (NT), neuroplasticity and drug efficacy (NP, EFF), and pharmacokinetics and hormonal regulation (PK, H). The non-CNS, peripheral system pharmacokinetic (SPK) subnetwork of the ketamine pharmacogenomic network could not be determined from the output of this system.

[0213] Figure 50 The lamotrigine pharmacogenomic adverse event subnetwork was shown, and post-hoc bioinformatic analysis showed that the lamotrigine adverse event subnetwork was significantly associated with Steven Johnson syndrome, drug-induced hypersensitivity, progressive cognitive impairment, choreiform movements, and headaches with dizziness.

[0214] Figure 51 The lamotrigine pharmacogenomic neuroplasticity subnetwork was shown, and post-hoc bioinformatic analysis showed that the lamotrigine pharmacogenomic neuroplasticity subnetwork was significantly associated with development and neurogenesis of neurons and disease states including epilepsy, fibromyalgia, bipolar 1 disorder, and mania.

[0215] Figure 52 is a list of all genes contained within the chromatin remodeling (CR) subnetwork of the lamotrigine pharmacogenomic network in the human brain.

[0216] Figure 53-1 and 53-2 is a list of all genes contained within the neuroplasticity and efficacy (NP, EFF) subnetwork of the lamotrigine pharmacogenomic network in the human brain.

[0217] Figure 54 is a list of all genes contained within the adverse event (AE) subnetwork of the lamotrigine pharmacogenomic network in the human brain.

[0218] Figure 55 is a list of all genes contained in the pharmacokinetic (PK) subnetwork of the lamotrigine pharmacogenomic network in the human brain.

[0219] Figure 56 shows the results of post-hoc bioinformatic analysis of the clozapine pharmacogenomic network and its accompanying network topological model. Figure 56A The most significant disease annotations for genes contained in the clozapine pharmacogenomic network were shown to be psychosis, agitation, bipolar spectrum disorder, nonaffective psychosis, and treatment-resistant schizophrenia.

[0220] Figure 56B The most significant drugs acting as upstream regulators of the clozapine pharmacogenomic network were shown to include clozapine (p-value = 8.85E-110; Fisher's exact test), haloperidol (p = 1.45E-42), and chloφromazine (p = 6.95E-20).

[0221] Figure 56C The clozapine pharmacogenomic network was shown to fit model network topological label 3, where the disassembled gene set subnetworks include chromatin remodeling (CR), adverse events, central nervous system (AE), pharmacokinetics (PK), and adverse events, peripheral immune system (IAE). The non-CNS, peripheral system pharmacokinetics (SPK) subnetwork of the clozapine pharmacogenomic network could not be determined from the output of this system.

[0222] Figure 57 The clozapine pharmacogenomic adverse event CNS and peripheral immune system adverse event subnetworks (AE, IAE) are shown, and post-hoc bioinformatic analysis showed that the clozapine adverse event subnetworks were significantly associated with carbohydrate metabolism disorders, systemic autoimmune disorders, weight gain, drug-induced neutropenia, and apoptosis of neurons.

[0223] Figure 58 The clozapine pharmacogenomic efficacy subnetwork (EFF) is shown, and post-hoc bioinformatic analysis showed that the clozapine efficacy subnetwork was significantly associated with psychosis, treatment-resistant schizophrenia, nonaffective psychosis, manic bipolar disorder, recurrent schizophrenia, mania, bipolar disorder, refractory schizophrenia, and schizophrenia.

[0224] Figure 59 is a list of all genes contained in the chromatin remodeling (CR) subnetwork of the clozapine pharmacogenomic network in the human brain.

[0225] Figure 60-1 and 60-2 is a list of all genes contained in the efficacy (EFF) subnetwork of the clozapine pharmacogenomic network in the human brain.

[0226] Figure 61-1 and 61-2 is a list of all genes contained in the adverse event (AE) subnetwork of the clozapine pharmacogenomic network in the human brain and in the peripheral immune system.

[0227] Figure 62 is a list of all genes contained in the pharmacokinetic (PK) subnetwork of the clozapine pharmacogenomic network in the human brain.

[0228] Another application of these methods is to identify genes that encode novel druggable molecules that are part of a drug efficacy subnetwork for which the function is known but the class of drug to which it belongs is not known. As shown in Figure 63, the warfarin pharmacogenomic network contains several candidate drug targets that were not previously known to be members of the anticoagulant pharmacogenomic network for this drug. The method of pharmacogenomic network mapping using SNPs as part of drug pathway reconstruction enables the addition of the following genes that mediate the anticoagulant effects of warfarin as part of subnetwork 1 : AXL (AXL receptor tyrosine kinase), F9 (coagulation factor IX), MERTK (MER proto-oncogene, tyrosine kinase), PDGFB (platelet-derived growth factor subunit B), PROC (protein C, inactivator of coagulation factors Va and VIIIa), PROCR (protein C receptor), PROS1 (protein S), and PROZ (protein Z, vitamin K-dependent). Some of these novel genes can encode druggable products for use as anticoagulants.

[0229] As shown in Figure 63, the identified pharmacogenomic network for warfarin contains one or more of the following: ABO, alpha

[0230] 1-3-N-acetylgalactosaminyltransferase and alpha 1-3-galactosyltransferase (ABO) gene, aldo-keto reductase family 1 member C3 (AKR1C3) gene, AXL receptor tyrosine kinase (AXL) gene, complement factor H-related 5 (CFHR5) gene, cytochrome P450 family 2, subfamily C, member 19 (CYP2C19) gene, cytochrome P450 family 2, subfamily C, member 8 (CYP2C8) gene, cytochrome P450 family 2, subfamily member C, member 9 (CYP2C9) gene, cytochrome P450 family 3, subfamily A, member 4 (CYP3A4) gene, cytochrome P450 family 4, subfamily F, member 2 (CYP4F2) gene, erythropoietin (EPO) gene, coagulation factor V (F5) gene, coagulation factor VII (F7) gene, coagulation factor IX (F9) gene, coagulation factor X (F10) gene, coagulation factor XI (F11) gene, coagulation factor XII (F12) gene, coagulation factor XIII A chain (F13A1) gene, fibrillin alpha chain (FGA) gene, fibrillin gamma chain (FGG) gene, growth arrest-specific 6 (GAS6) gene, histidine-rich glycoprotein (HRG) gene, kininogen 1 (KNG1) gene, lysozyme (LYZ) gene, MER proto-oncogene, tyrosine kinase (MERTK) gene, matrix Gla protein (MGP) gene, orosomucoid 1 (ORM1) gene, polycomb group ring finger 3 (PCGF3) gene, platelet-derived growth factor subunit B (PDGFB) gene, protein C, coagulation factors Va and VIIIa inactivator (PROC) gene, protein C receptor (PROCR) gene, protein S (PROS1) gene, protein Z, vitamin K-dependent plasma glycoprotein (PROZ) gene, serine protease 8 (PRSS8) gene, serine protease 53 (PRSS53) gene, sphingosine kinase 1 (SPHK1) gene, signal transducer and activator of transcription 3 (STAT3) gene, synaptophysin 4 (STX4) gene, surfeit 4 (SURF4) gene, transient receptor potential cation channel subfamily C member 4 associated protein (TRPC4AP) gene, ubiquitin specific peptidase 7 (USP7) gene, vitamin K epoxide reductase complex subunit 1 (VKORC1) gene, vitamin K epoxide reductase complex subunit 1 like 1 (VKORC1L1) gene, or von Willebrand factor (VWF) gene.

[0231] Throughout this specification, plural instances can implement components, operations, or structures described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations can be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components in example configurations can be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component can be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter of this document.

[0232] Additionally, certain embodiments are described as comprising, consisting of, consisting essentially of, or as including particular components, elements, or steps. The term "comprising" is used herein to mean that the named component(s) is included, but not excluding the presence of one or more other components, steps, or elements. The term "consisting essentially of" is used herein to mean that the named component(s) is included, and that one or more other components, steps, or elements are not excluded, but that the composition or method does not include additional components, steps, or elements. The term "consisting of" is used herein to mean that the named component(s) is included, and that one or more other components, steps, or elements are excluded.

[0233] In various embodiments, a hardware module can be implemented mechanically or electronically. For example, a hardware module can include dedicated circuitry or logic that is permanently configured to perform certain operations. Such a hardware module can be specifically configured hardware modules or programmable hardware modules. A programmable hardware module can be configured by one or more software applications or firmware applications stored on or running on a processor. For example, a general purpose processor can be programmed to become a special-purpose processor in the execution of software or firmware applications. A hardware module can also include programmable logic or circuitry that is temporarily configured by software or firmware applications.

[0234] Accordingly, the term "hardware module" should be understood to encompass a tangible entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering embodiments in which hardware modules are temporarily configured (e.g., programmed), each of the hardware modules need not be configured or instantiated at any one instance in time. For example, where the hardware modules comprise a general-purpose processor configured by software to become a special-purpose processor, the general-purpose processor can be configured as

[0235] Hardware modules can provide information to, and receive information from, other hardware modules. Accordingly, the hardware modules can be regarded as being communicatively coupled. Where multiple of such hardware modules exist contemporaneously, communications can be achieved through signal transmission (e.g., over appropriate circuits and buses) that connect the hardware modules. In embodiments in which multiple hardware modules are configured or instantiated at different times, communications between such hardware modules can be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware modules have access. For example, one hardware module can perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware module can then, at a later time, access the memory device to retrieve and process the stored output. Hardware modules can also initiate communications with input or output devices, and can operate on a resource (e.g., a collection of information).

[0236] The various operations of example methods described herein can be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors can constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may, in some example embodiments, comprise processor-implemented modules.

[0237] Similarly, the methods or routines described herein can be at least partially processor- implemented. For example, at least some of the operations of a method can be performed by one or more processors or processor-implemented hardware modules. The performance of certain of the operations can be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the processor or processors can be located in a single location (e.g., within a home environment, an office environment, or as a server farm), while in other embodiments the processors can be distributed across many geographic locations.

[0238] The performance of certain of the operations can be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the one or more processors or processor-implemented modules can be located in a single geographic location (e.g., within a home environment, an office environment, or a server farm). In other embodiments, the one or more processors or processor-implemented modules can be distributed across a number of geographic locations.

[0239] Unless specifically stated otherwise, discussions herein using terms such as "processing," "computing," "calculating," "determining," "presenting," "displaying," or the like can refer to actions or processes of a machine (e.g., a computer) that manipulates or transforms data represented as physical (e.g., electronic) quantities within one or more memories (e.g., volatile memory, non-volatile memory, or a combination thereof), registers or other machine component(s) that receive, store, communicate, or display information.

[0240] As used herein, any reference to "one implementation" or "an implementation" means that a particular element, feature, structure, or characteristic described in connection with the implementation is included in at least one implementation. The appearances of the phrase "in one implementation" in various places in the specification are not necessarily all referring to the same implementation.

[0241] Some embodiments can be described using the expression "coupled" and "connected" along with their derivatives. For example, some embodiments can be described using the term "coupled" to refer to two or more elements being in direct physical or electrical contact with one another. The term "coupled," however, can also mean that two or more elements are not in direct contact with one another, but yet still co-operate or interact with one another. Embodiments are not limited to these specific types of relationships.

[0242] As used herein, the terms "comprises," "comprising," "includes," "including," "has," "having" or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, article or apparatus that comprises a list of elements is not necessarily limited to only those elements but can include other elements not expressly listed or inherent to such process, method, article or apparatus. Further, unless expressly stated to the contrary, "or" refers to an inclusive or and not to an exclusive or. For example, a condition A or B is satisfied by any one of the following: A is true (or present) and B is false (or not present), A is false (or not present) and B is true (or present), and some combination thereof.

[0243] Additionally, "a" or "an" can be used to describe elements and components of the embodiments herein. This is merely for convenience and to give a general sense of the reference in which it is used. This description and the accompanying claims should be read to include one or at least one and the singular word should not be used to the exclusion of the plural unless specifically so stated.

[0244] This detailed description should be construed to cover any alternatives falling within the equivalent constructive description. The drawings shown in this patent document are included simply to provide a conceptual understanding of the structures being discussed. The drawings are not intended to be precise in their depiction of the structures.

Claims

1. A computer-implemented method for identifying a pharmacogenomic network of a drug, the method performed by one or more processors programmed to perform the method, the method comprising: obtaining, by one or more processors, a plurality of single nucleotide polymorphisms (SNPs) associated with a drug response or adverse event of a particular drug; identifying, by the one or more processors, a pharmacogenomic network of the particular drug based on a gene set associated with the plurality of SNPs; deconstructing, by the one or more processors, the pharmacogenomic network into a plurality of pharmacogenomic subnetworks using a subnetwork type topology selected from two or more of a chromatin remodeling subnetwork type, a drug efficacy subnetwork type, a drug adverse event subnetwork type, a pharmacokinetic enzyme and hormone subnetwork type, a systemic pharmacokinetic subnetwork type, or an immune system adverse event subnetwork type, using iterative gene set optimization to organize the gene set into functional subsets to identify a plurality of pharmacogenomic subnetworks of the particular drug corresponding to each of the functional subsets; and providing, by the one or more processors, for display an indication of the pharmacogenomic network and an indication of the plurality of pharmacogenomic subnetworks.

2. The method of claim 1, wherein identifying a pharmacogenomic network of the particular drug based on a gene set associated with the plurality of SNPs comprises: comparing, by the one or more processors, the plurality of SNPs to a SNP database using topologically associated domain (TAD) boundaries within a chromosome domain to identify additional SNPs connected to the plurality of SNPs, wherein the plurality of SNPs and the additional SNPs are included in a candidate variant set; performing, by the one or more processors, a mapping of 3D spatial connections using the candidate variant set as probes within chromatin data to determine interconnections between target genes associated with a drug response or adverse event of the particular drug; performing, by the one or more processors, pathway analysis on the target genes associated with the candidate variant set to filter the target genes to identify a gene set causally relevant to the particular drug; and identifying, by the one or more processors, the pharmacogenomic network of the particular drug based on the identified gene set.

3. The method of claim 2, further comprising: performing, by the one or more processors, bioinformatics analysis on the pharmacogenomic network to ensure that the most significantly associated drug with the plurality of pharmacogenomic subnetworks is the particular drug.

4. The method of claim 3, wherein performing bioinformatics analysis on the pharmacogenomic network comprises: validating, by bioinformatics, the pharmacogenomic network and the plurality of pharmacogenomic subnetworks of the particular drug.

5. The method of claim 4, wherein validating the drug pharmacogenomic network and the plurality of drug pharmacogenomic sub-networks of the particular drug comprises: comparing the drug pharmacogenomic network and the plurality of drug pharmacogenomic sub-networks of the particular drug to one or more of: terms from a gene ontology or a drug database; canonical biological pathways in a cell or tissue in which the particular drug acts, or dysbiotic upstream regulators in the cell or tissue in which the particular drug acts.

6. The method of claim 2, wherein performing pathway analysis on the target genes associated with the subset of intermediate candidate variants to filter the target genes comprises identifying, by the one or more processors, a candidate drug pharmacogenomic network gene set by identifying a subset of the target genes that form statistically significant interconnected pathways expressed in a tissue associated with the particular drug.

7. The method of claim 6, further comprising: analyzing each gene in the candidate drug pharmacogenomic network gene set according to at least one of: a function of the gene in the context of the particular drug, a set of mutations within the gene, or a pattern of expression of the gene relative to a neuroanatomical substrate as defined by RNA expression data, functional imaging, or other integrated multi-scale data indicative of where the particular drug is known to act; analyzing each gene in the candidate drug pharmacogenomic network gene set based on other genes and functional genomic elements including long non-coding RNAs; and adding or removing genes from the candidate drug pharmacogenomic network gene set based on the analysis.

8. The method of claim 7, wherein analyzing each gene in the candidate drug pharmacogenomic network gene set according to a pattern of expression of the gene relative to a neuroanatomical substrate comprises: comparing each gene in the candidate drug pharmacogenomic network gene set to a neurological map indicative of the neuroanatomical substrate of the particular drug; and filtering from the candidate drug pharmacogenomic network gene set genes that are not expressed in the same neuroanatomical region as the particular drug.

9. The method of any one of claims 1-8, further comprising: determining, by the one or more processors, one or more spatial contacts comprising TADs based on the plurality of single nucleotide polymorphisms (SNPs) associated with a drug response or adverse event of the particular drug, wherein the one or more spatial contacts are differentially expanded and repressed.

10. The method of any one of claims 1-8, wherein the drug pharmacogenomic network comprises one of more defining properties of functional spatial genomics, the more defining properties comprising at least one of: a mutation affecting an enhancer-promoter pair; a promoter-promoter pair; a super-enhancer-promoter pair; a regulatory RNA; an euchromatin or heterochromatin state; a topologically associated domain (TAD); or a chromatin state. Lamina-associated domain (LAD).

11. The method of any one of claims 1-8, further comprising: storing, by the one or more processors, the indication of the drug pharmacogenomic network for the particular drug and the indications of the plurality of drug pharmacogenomic subnetworks in a database as a reference set to compare and score an input patient pharmacogenomic network for the same particular drug to determine efficacy and adverse events in treating a condition or disease of the patient.

12. The method of any one of claims 1-8, further comprising: identifying, by the one or more processors, a set of molecular pharmacodynamic targets within one of the plurality of drug pharmacogenomic subnetworks for the same or similar clinical indication as a specified drug.

13. The method of claim 12, wherein the set of molecular pharmacodynamic targets within one of the plurality of drug pharmacogenomic subnetworks and the particular drug act in the human brain and a majority of members of the pathway of the molecular pharmacodynamic targets are expressed in the same region of the human brain as the particular drug.

14. The method of any one of claims 1-8, further comprising: re-purposing drugs based on potential loci of pharmacogenomic enhancer SNPs related to the plurality of drug pharmacogenomic subnetworks within the topology of the subnetwork type.

15. The method of any one of claims 1-8, further comprising: re-purposing a combination of drugs to be tested as a therapeutic agent after determining that physiological mechanisms exhibit complementarity between the subnetwork properties of two or more drugs provides a superior therapy for a particular disease or disease state than a single particular drug.

16. The method of claim 15, wherein re-purposing a combination of drugs comprises re-purposing valproic acid and ketamine as a combination therapeutic agent for neurological and neuropsychiatric disorders.

17. The method of any one of claims 1-7, wherein the particular drug is one of: valproic acid, ketamine, lithium, lamotrigine, clozapine, or warfarin.

18. A computing device for identifying a pharmacogenomic network of a drug, the computing device comprising: a communication network; one or more processors; and a non-transitory computer-readable memory coupled to the one or more processors and storing instructions thereon that, when executed by the one or more processors, cause the computing device to perform the steps of the method of any one of claims 1-17.

Images