TransGEM: a molecule generation model based on transformer with gene expression data.

MOTIVATION: It is difficult to generate new molecules with desirable bioactivity through ligand-based de novo drug design, and receptor-based de novo drug design is constrained by disease target information availability. The combination of artificial intelligence and phenotype-based de novo drug design can generate new bioactive molecules, independent from disease target information. Gene expression profiles can be used to characterize biological phenotypes. The Transformer model can be utilized to capture the associations between gene expression profiles and molecular structures due to its remarkable ability in processing contextual information. RESULTS: We propose TransGEM (Transformer-based model from gene expression to molecules), which is a phenotype-based de novo drug design model. A specialized gene expression encoder is employed to embed gene expression difference values between diseased cell lines and their corresponding normal tissue cells into TransGEM model. The results demonstrate that the TransGEM model can generate molecules with desirable evaluation metrics and property distributions. Case studies illustrate that TransGEM model can generate structurally novel molecules with good binding affinity to disease target proteins. The majority of genes with high attention scores obtained from TransGEM model are associated with the onset of the disease, indicating the potential of these genes as disease targets. Therefore, this study provides a new paradigm for de novo drug design, and it will promote phenotype-based drug discovery. AVAILABILITY: The code is available at https://github.com/hzauzqy/TransGEM. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Glycolysis-Stimulated Esrrb Lactylation Promotes the Self-Renewal and Extraembryonic Endoderm Stem Cell Differentiation of Embryonic Stem Cells.

Embryonic stem cells (ESCs) favor glycolysis over oxidative phosphorylation for energy production, and glycolytic metabolism is critical for pluripotency establishment, maintenance, and exit. However, an understanding of how glycolysis regulates the self-renewal and differentiation of ESCs remains elusive. Here, we demonstrated that protein lactylation, regulated by intracellular lactate, contributes to the self-renewal of ESCs. We further showed that Esrrb, an orphan nuclear receptor involved in pluripotency maintenance and extraembryonic endoderm stem cell (XEN) differentiation, is lactylated on K228 and K232. The lactylation of Esrrb enhances its activity in promoting ESC self-renewal in the absence of the LIF and XEN differentiation of ESCs by increasing its binding at target genes. Our studies reveal the importance of protein lactylation in the self-renewal and XEN differentiation of ESCs, and the underlying mechanism of glycolytic metabolism regulating cell fate choice.

Metabolites of pathogenic microorganisms database (MPMdb) and its seed metabolite applications.

Seed metabolites are the combination of essential compounds required by an organism across various potential environmental conditions. The seed metabolites screening framework based on the network topology approach can capture important biological information of species. This study aims to identify comprehensively the relationship between seed metabolites and pathogenic bacteria. A large-scale data set was compiled, describing the seed metabolite sets and metabolite sets of 124,192 pathogenic strains from 34 genera, by constructing genome-scale metabolic models. The enrichment analysis method was used to screen the specific seed metabolites of each species/genus of pathogenic bacteria. The metabolites of pathogenic microorganisms database (MPMdb) (http://qyzhanglab.hzau.edu.cn/MPMdb/) was established for browsing, searching, predicting, or downloading metabolites and seed metabolites of pathogenic microorganisms. Based on the MPMdb, taxonomic and phylogenetic analyses of pathogenic bacteria were performed according to the function of seed metabolites and metabolites. The results showed that the seed metabolites could be used as a feature for microorganism chemotaxonomy, and they could mirror the phylogeny of pathogenic bacteria. In addition, our screened specific seed metabolites of pathogenic bacteria can be used not only for further tapping the nutritional resources and identifying auxotrophies of pathogenic bacteria but also for designing targeted bactericidal compounds by combining with existing antimicrobial agents.IMPORTANCEMetabolites serve as key communication links between pathogenic microorganisms and hosts, with seed metabolites being crucial for microbial growth, reproduction, external communication, and host infection. However, the large-scale screening of metabolites and the identification of seed metabolites have always been the main technical bottleneck due to the low throughput and costly analysis. Genome-scale metabolic models have become a recognized research paradigm to investigate the metabolic characteristics of species. The developed metabolites of pathogenic microorganisms database in this study is committed to systematically predicting and identifying the metabolites and seed metabolites of pathogenic microorganisms, which could provide a powerful resource platform for pathogenic bacteria research.

The Role of Antisolvents with Different Functional Groups in the Formation of Cs4PbBr6 and CsPbBr3 Particles.

Compared to the high-temperature hot injection (HI) technique, the room-temperature supersaturated recrystallization (SR) approach is more hopeful to realize the industrialized production of CsPbX3 (X = Cl, Br, and I) nanomaterials. However, accurate compositional control of the product is still difficult, and the role and underlying mechanism of antisolvents in the reprecipitation process remain unclear. Herein, CsPbBr3 particles and CsPbBr3/Cs4PbBr6 composites with certain proportions are synthesized using different antisolvents with the SR method. By adjustment of the polarity or functional group of antisolvents, it is found that the functional groups of antisolvents have a major impact on the composition of the products. Furthermore, the influential mechanism of different antisolvents on the compositions of products is investigated by combining electrostatic potential calculations and ultraviolet-visible absorption spectroscopy. It suggests that the interaction between functional groups of antisolvents and organic ligands influences the coordination status of the intermediate Pb-complex and further affects the separating rate of the Pb(II)-intermediate, leading to the formation of products with different compositions. A physicochemical mechanism is proposed to explain the formation of Cs4PbBr6 and CsPbBr3. This work deepens the understanding of the formation mechanism of all-inorganic metal halide perovskite-related materials based on the SR method and provides new routes to achieve their controllable preparation.

Characterization of brain resilience in Alzheimer's disease using polygenic risk scores and further improvement by integrating mitochondria-associated loci.

INTRODUCTION: Identification of high-risk people for Alzheimer's disease (AD) is critical for prognosis and early management. Longitudinal epidemiologic studies have observed heterogeneity in the brain and cognitive aging. Brain resilience was described as above-expected cognitive function. The "resilience" framework has been shown to correlate with individual characteristics such as genetic factors and age. Besides, accumulative evidence has confirmed the association of mitochondria with the pathogenesis of AD. However, it is challenging to assess resilience through genetic metrics, in particular incorporating mitochondria-associated loci. OBJECTIVES: In this paper, we first demonstrated that polygenic risk scores (PRS) could characterize individuals' resilience levels. Then, we indicated that mitochondria-associated loci could improve the performance of PRSs, providing more reliable measurements for the prevention and diagnosis of AD. METHODS: The discovery (N = 1,550) and independent validation samples (N = 2,090) were used to construct nine types of PRSs containing mitochondria-related loci (PRSMT) from both biological and statistical aspects and combined them with known AD risk loci derived from genome-wide association studies (GWAS).Individuals' levels of brain resilience were comprehensively measured by linear regression models using eight pathological characteristics. RESULTS: It was found that PRSs could characterize brain resilience levels (e.g., Pearson correlation test Pmin = 7.96×10-9). Moreover, the performance of PRS models could be efficiently improved by incorporating a small number of mitochondria-related loci (e.g., Pearson correlation test P improved from 1.41×10-3 to 6.09×10-6). PRSs' ability to characterize brain resilience was validated. More importantly, by incorporating some mitochondria-related loci, the performance of PRSs in measuring brain resilience could be significantly improved. CONCLUSION: Our findings imply that mitochondria may play an important role in brain resilience, and targeting mitochondria may open a new door to AD prevention and therapy.

Dissecting Mitochondrial Mechanisms of Alzheimer's Disease Using Gene Dependency Network and Its Implications for Discovering Nutrients Combatting the Disease.

BACKGROUND: Alzheimer's disease (AD) is the leading cause of dementia, with its prevalence increasing as the global population ages. AD is a multifactorial and intricate neurodegenerative disease with pathological changes varying from person to person. Because the mechanism of AD is highly controversial, effective treatments remain a distant prospect. Currently, one of the most promising hypotheses posits mitochondrial dysfunction as an early event in AD diagnosis and a potential therapeutic target. OBJECTIVE: Here, we adopted a systems medicine strategy to explore the mitochondria-related mechanisms of AD. Then, its implications for discovering nutrients combatting the disease were demonstrated. METHODS: We employed conditional mutual information (CMI) to construct AD gene dependency networks. Furthermore, the GeneRank algorithm was applied to prioritize the gene importance of AD patients and identify potential anti-AD nutrients targeting crucial genes. RESULTS: The results suggested that two highly interconnected networks of mitochondrial ribosomal proteins (MRPs) play an important role in the regulation of AD pathology. The close association between mitochondrial ribosome dysfunction and AD was identified. Additionally, we proposed seven nutrients with potential preventive and ameliorative effects on AD, five of which have been supported by experimental reports. CONCLUSIONS: Our study explored the important regulatory role of MRP genes in AD, which has significant implications for AD prevention and treatment.

An integrated method for studying the biodegradation of benzo[a]pyrene by Citrobacter sp. HJS-1 and interaction mechanism based on the structural model of the initial dioxygenase.

A bacterial strain Citrobacter sp. HJS-1 was discovered from the sludge in a drainage canal of a coal mine. Firstly, its biodegradation capacity for benzo[a]pyrene (BaP) was detected under different concentrations. The results proved that the strain possessed excellent biodegradation capacity for BaP with high-efficiency degradation rates ranging from 78.9 to 86.8%. The highest degradation rate was observed in the low-concentration sample, and the high-concentration BaP had a slight influence on the biodegradation capacity due to the potential toxicity of BaP and its oxygen-containing derivatives. Meanwhile, the degradation test for the other five aromatic hydrocarbons (2- to 4-ring) proved that the strain had a comprehensive degradation potential. To clarify the biodegradation mechanism of BaP, a dioxygenase structure was constructed by homology modeling. Then, the interactions between dioxygenase and BaP were researched by molecular simulation. Combined with the identification of the vital BaP-cis-7,8-dihydrodiol intermediate and the interaction analysis, the initial oxidation mode and the binding site of BaP were revealed in the dioxygenase. Taken together, this study has offered a way to understand the biodegradation process of BaP and its interaction mechanism based on experimental and theoretical analysis.

Biotransformation of benzo[a]pyrene by Pannonibacter sp. JPA3 and the degradation mechanism through the initially oxidized benzo[a]pyrene-4,5-dihydrodiol to downstream metabolites.

Owing to its adverse effects on the environment and human health, benzo[a]pyrene (BaP) has attracted considerable attention and has been used as a model compound in ecotoxicology. In this study, Pannonibacter sp. JPA3 as a BaP-degrading strain was isolated from the production water of an oil well. The strain could remove 80% of BaP at an initial concentration of 100 mg L-1 after 35 d culture. The BaP-4,5-dihydrodiol, BaP-4,5-epoxide, 5-hydroxychrysene, and 2-hydroxy-1-naphthoic acid metabolites were identified in the biodegradation process. Simultaneously, the gene sequence coding for dioxygenase in the strain was amplified and a dioxygenase model was built by homology modeling. Combined with the identification of the metabolites, the interaction mechanism of BaP with dioxygenase was investigated using molecular docking. It was assumed that BaP was initially oxidized at the C4-C5 positions in the active cavity of dioxygenase. Moreover, a hypothesis for the progressive degradation mechanism of BaP by this strain was proposed via the identification of the downstream metabolites. In conclusion, our study provided an efficient BaP degrader and a comprehensive reference for the study of the degradation mechanism in terms of the degrading metabolites and theoretical research at the molecular level.

Transporter proteins knowledge graph construction and its application in drug development.

Transporters are the main determinant for pharmacokinetics characteristics of drugs, such as absorption, distribution, and excretion of drugs in humans. However, it is difficult to perform drug transporter validation and structure analysis of membrane transporter proteins by experimental methods. Many studies have demonstrated that knowledge graphs (KG) could effectively excavate potential association information between different entities. To improve the effectiveness of drug discovery, a transporter-related KG was constructed in this study. Meanwhile, a predictive frame (AutoInt_KG) and a generative frame (MolGPT_KG) were established based on the heterogeneity information obtained from the transporter-related KG by the RESCAL model. Natural product Luteolin with known transporters was selected to verify the reliability of the AutoInt_KG frame, its ROC-AUC (1:1), ROC-AUC (1:10), PR-AUC (1:1), PR-AUC (1:10) are 0.91, 0.94, 0.91 and 0.78, respectively. Subsequently, the MolGPT_KG frame was constructed to implement efficient drug design based on transporter structure. The evaluation results showed that the MolGPT_KG could generate novel and valid molecules and that these molecules were further confirmed by molecular docking analysis. The docking results showed that they could bind to important amino acids at the active site of the target transporter. Our findings will provide rich information resources and guidance for the further development of the transporter-related drugs.

Evolution-strengthened knowledge graph enables predicting the targetability and druggability of genes.

Identifying promising targets is a critical step in modern drug discovery, with causative genes of diseases that are an important source of successful targets. Previous studies have found that the pathogeneses of various diseases are closely related to the evolutionary events of organisms. Accordingly, evolutionary knowledge can facilitate the prediction of causative genes and further accelerate target identification. With the development of modern biotechnology, massive biomedical data have been accumulated, and knowledge graphs (KGs) have emerged as a powerful approach for integrating and utilizing vast amounts of data. In this study, we constructed an evolution-strengthened knowledge graph (ESKG) and validated applications of ESKG in the identification of causative genes. More importantly, we developed an ESKG-based machine learning model named GraphEvo, which can effectively predict the targetability and the druggability of genes. We further investigated the explainability of the ESKG in druggability prediction by dissecting the evolutionary hallmarks of successful targets. Our study highlights the importance of evolutionary knowledge in biomedical research and demonstrates the potential power of ESKG in promising target identification. The data set of ESKG and the code of GraphEvo can be downloaded from https://github.com/Zhankun-Xiong/GraphEvo.

Reconstructed Genome-Scale Metabolic Model Characterizes Adaptive Metabolic Flux Changes in Peripheral Blood Mononuclear Cells in Severe COVID-19 Patients.

Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) poses a mortal threat to human health. The elucidation of the relationship between peripheral immune cells and the development of inflammation is essential for revealing the pathogenic mechanism of COVID-19 and developing related antiviral drugs. The immune cell metabolism-targeting therapies exhibit a desirable anti-inflammatory effect in some treatment cases. In this study, based on differentially expressed gene (DEG) analysis, a genome-scale metabolic model (GSMM) was reconstructed by integrating transcriptome data to characterize the adaptive metabolic changes in peripheral blood mononuclear cells (PBMCs) in severe COVID-19 patients. Differential flux analysis revealed that metabolic changes such as enhanced aerobic glycolysis, impaired oxidative phosphorylation, fluctuating biogenesis of lipids, vitamins (folate and retinol), and nucleotides played important roles in the inflammation adaptation of PBMCs. Moreover, the main metabolic enzymes such as the solute carrier (SLC) family 2 member 3 (SLC2A3) and fatty acid synthase (FASN), responsible for the reactions with large differential fluxes, were identified as potential therapeutic targets. Our results revealed the inflammation regulation potentials of partial metabolic reactions with differential fluxes and their metabolites. This study provides a reference for developing potential PBMC metabolism-targeting therapy strategies against COVID-19.

Systems Genetic Identification of Mitochondrion-Associated Alzheimer's Disease Genes and Implications for Disease Risk Prediction.

Cumulative evidence has revealed the association between mitochondrial dysfunction and Alzheimer's disease (AD). Because the number of mitochondrial genes is very limited, the mitochondrial pathogenesis of AD must involve certain nuclear genes. In this study, we employed systems genetic methods to identify mitochondrion-associated nuclear genes that may participate in the pathogenesis of AD. First, we performed a mitochondrial genome-wide association study (MiWAS, n = 809) to identify mitochondrial single-nucleotide polymorphisms (MT-SNPs) associated with AD. Then, epistasis analysis was performed to examine interacting SNPs between the mitochondrial and nuclear genomes. Weighted co-expression network analysis (WGCNA) was applied to transcriptomic data from the same sample (n = 743) to identify AD-related gene modules, which were further enriched by mitochondrion-associated genes. Using hub genes derived from these modules, random forest models were constructed to predict AD risk in four independent datasets (n = 743, n = 542, n = 161, and n = 540). In total, 9 potentially significant MT-SNPs and 14,340 nominally significant MT-nuclear interactive SNPs were identified for AD, which were validated by functional analysis. A total of 6 mitochondrion-related modules involved in AD pathogenesis were found by WGCNA, from which 91 hub genes were screened and used to build AD risk prediction models. For the four independent datasets, these models perform better than those derived from AD genes identified by genome-wide association studies (GWASs) or differential expression analysis (DeLong's test, p < 0.05). Overall, through systems genetics analyses, mitochondrion-associated SNPs/genes with potential roles in AD pathogenesis were identified and preliminarily validated, illustrating the power of mitochondrial genetics in AD pathogenesis elucidation and risk prediction.

Hbxip is essential for murine embryogenesis and regulates embryonic stem cell differentiation through activating mTORC1.

HBXIP, also named LAMTOR5, has been well characterized as a transcriptional co-activator in various cancers. However, the role of Hbxip in normal development remains unexplored. Here, we demonstrated that homozygous knockout of Hbxip leads to embryonic lethality, with retarded growth around E7.5, and that depletion of Hbxip compromises the self-renewal of embryonic stem cells (ESCs), with reduced expression of pluripotency genes, reduced cell proliferation and decreased colony-forming capacity. In addition, both Hbxip-/- ESCs and E7.5 embryos displayed defects in ectodermal and mesodermal differentiation. Mechanistically, Hbxip interacts with other components of the Ragulator complex, which is required for mTORC1 activation by amino acids. Importantly, ESCs depleted of Ragulator subunits, Lamtor3 or Lamtor4, displayed differentiation defects similar to those of Hbxip-/- ESCs. Moreover, Hbxip-/-, p14-/- and p18-/- mice, lacking subunits of the Ragulator complex, also shared similar phenotypes, embryonic lethality and retarded growth around E7-E8. Thus, we conclude that Hbxip plays a pivotal role in the development and differentiation of the epiblast, as well as the self-renewal and differentiation of ESCs, through activating mTORC1 signaling.

Agent Clustering Strategy Based on Metabolic Flux Distribution and Transcriptome Expression for Novel Drug Development.

The network module-based method has been used for drug repositioning. The traditional drug repositioning method only uses the gene characteristics of the drug but ignores the drug-triggered metabolic changes. The metabolic network systematically characterizes the connection between genes, proteins, and metabolic reactions. The differential metabolic flux distribution, as drug metabolism characteristics, was employed to cluster the agents with similar MoAs (mechanism of action). In this study, agents with the same pharmacology were clustered into one group, and a total of 1309 agents from the CMap database were clustered into 98 groups based on differential metabolic flux distribution. Transcription factor (TF) enrichment analysis revealed the agents in the same group (such as group 7 and group 26) were confirmed to have similar MoAs. Through this agent clustering strategy, the candidate drugs which can inhibit (Japanese encephalitis virus) JEV infection were identified. This study provides new insights into drug repositioning and their MoAs.

Facilitating Antiviral Drug Discovery Using Genetic and Evolutionary Knowledge.

Over the course of human history, billions of people worldwide have been infected by various viruses. Despite rapid progress in the development of biomedical techniques, it is still a significant challenge to find promising new antiviral targets and drugs. In the past, antiviral drugs mainly targeted viral proteins when they were used as part of treatment strategies. Since the virus mutation rate is much faster than that of the host, such drugs feature drug resistance and narrow-spectrum antiviral problems. Therefore, the targeting of host molecules has gradually become an important area of research for the development of antiviral drugs. In recent years, rapid advances in high-throughput sequencing techniques have enabled numerous genetic studies (such as genome-wide association studies (GWAS), clustered regularly interspersed short palindromic repeats (CRISPR) screening, etc.) for human diseases, providing valuable genetic and evolutionary resources. Furthermore, it has been revealed that successful drug targets exhibit similar genetic and evolutionary features, which are of great value in identifying promising drug targets and discovering new drugs. Considering these developments, in this article the authors propose a host-targeted antiviral drug discovery strategy based on knowledge of genetics and evolution. We first comprehensively summarized the genetic, subcellular location, and evolutionary features of the human genes that have been successfully used as antiviral targets. Next, the summarized features were used to screen novel druggable antiviral targets and to find potential antiviral drugs, in an attempt to promote the discovery of new antiviral drugs.

Self-Assembly of CsPbBr3 Nanocubes into 2D Nanosheets.

All-inorganic metal halide perovskites have attracted considerable attention due to their high application potentials in optoelectronics, photonics, and energy conversion. Herein, two-dimensional (2D) CsPbBr3 nanosheets with a thickness of about 3 nm have been synthesized through a simple chemical process based on a hot-injection technique. The lateral dimension of CsPbBr3 nanosheets ranges from 11 to 110 nm, which can be tuned by adjusting the ratio of short ligands (octanoic acid and octylamine) over long ligands (oleic acid and oleylamine). The nanosheets result from the self-assembly of CsPbBr3 nanocubes with an edge length of about 3 nm, which possess the same crystal orientation. In addition, an amorphous region of about 1 nm in width is found between adjacent nanocubes. To investigate both the structure and the growth mechanism of these nanosheets, microstructural characterizations at the atomic scale are conducted, combined with X-ray diffraction analysis, 1H nuclear magnetic resonance (1H NMR) measurement, and density functional theory (DFT) calculation, aiming to determine the configuration of different ligands adsorbed onto CsPbBr3. Our results suggest that the adjacent nanocubes are mainly connected together by short ligands and inclined long ligands. On the basis of the DFT calculation results, a relationship is derived for the volume ratio of short ligands over long ligands and the lateral dimensions of CsPbBr3 nanosheets. Moreover, a physicochemical mechanism is proposed to explain the 2D growth of CsPbBr3 nanosheets. Such a finding provides new insights regarding the well-ordered self-arrangement of CsPbBr3 nanomaterials, as well as new routes to synthesize 2D CsPbX3 (X = Cl and I) nanosheets of suitable dimensions for specific and large-scale applications.

Development of specific and selective bactericide by introducing exogenous metabolite of pathogenic bacteria.

The widespread and repeated use of broad-spectrum bactericides has led to an increase in resistance. Developing novel broad-spectrum bactericides cannot solve the resistance problem, and may even aggravate it. The design of specific and selective bactericides has become urgent. A specific bactericidal design strategy was proposed by introducing exogenous metabolites in this study. This strategy was used to optimize two known antibacterial agents, luteolin (M) and Isoprothiolane (D), against Xoo. Based on the prodrug principles, target compound MB and DB were synthesized by combing M or D with exogenous metabolites, respectively. Bactericidal activity test results demonstrated that while the antibacterial ability of target compounds was significantly improved, their selectivity was also well enhanced by the introducing of exogenous metabolites. Comparing with the original compound, the antibacterial activity of target compound was significantly increased 92.0% and 74.5%, respectively. The optimized target compounds were more easily absorbed, and the drug application concentrations were much lower than those of the original agents, which would greatly reduce environmental pollution and relieve resistance risk. Our proposed strategy is of great significance for exploring the specific and selective bactericides against other pathogens.

Co-immobilized Alcohol Dehydrogenase and Glucose Dehydrogenase with Resin Extraction for Continuous Production of Chiral Diaryl Alcohol.

Ni2+-functionalized porous ceramic/agarose composite beads (Ni-NTA Cerose) can be used as carrier materials to immobilize enzymes harboring a metal affinity tag. Here, a 6×His-tag fusion alcohol dehydrogenase Mu-S5 and glucose dehydrogenase from Bacillus megaterium (BmGDH) were co-immobilized on Ni-NTA Cerose to construct a packed bed reactor (PBR) for the continuous synthesis of the chiral intermediate (S)-(4-chlorophenyl)-(pyridin-2-yl) methanol ((S)-CPMA) NADPH recycling, and in situ product adsorption was achieved simultaneously by assembling a D101 macroporous resin column after the PBR. Using an optimum enzyme activity ratio of 2:1 (Mu-S5: BmGDH) and hydroxypropyl-ß-cyclodextrin as co-solvent, a space-time yield of 1560 g/(L·d) could be achieved in the first three days at a flow rate of 5 mL/min and substrate concentration of 10 mM. With simplified selective adsorption and extraction procedures, (S)-CPMA was obtained in 84% isolated yield.

Genome-wide epistasis analysis for Alzheimer's disease and implications for genetic risk prediction.

BACKGROUND: Single-nucleotide polymorphisms (SNPs) identified by genome-wide association studies only explain part of the heritability of Alzheimer's disease (AD). Epistasis has been considered as one of the main causes of "missing heritability" in AD. METHODS: We performed genome-wide epistasis screening (N = 10,389) for the clinical diagnosis of AD using three popularly adopted methods. Subsequent analyses were performed to eliminate spurious associations caused by possible confounding factors. Then, candidate genetic interactions were examined for their co-expression in the brains of AD patients and analyzed for their association with intermediate AD phenotypes. Moreover, a new approach was developed to compile the epistasis risk factors into an epistasis risk score (ERS) based on multifactor dimensional reduction. Two independent datasets were used to evaluate the feasibility of ERSs in AD risk prediction. RESULTS: We identified 2 candidate genetic interactions with PFDR <  0.05 (RAMP3-SEMA3A and NSMCE1-DGKE/C17orf67) and another 5 genetic interactions with PFDR <  0.1. Co-expression between the identified interactions supported the existence of possible biological interactions underlying the observed statistical significance. Further association of candidate interactions with intermediate phenotypes helps explain the mechanisms of neuropathological alterations involved in AD. Importantly, we found that ERSs can identify high-risk individuals showing earlier onset of AD. Combined risk scores of SNPs and SNP-SNP interactions showed slightly but steadily increased AUC in predicting the clinical status of AD. CONCLUSIONS: In summary, we performed a genome-wide epistasis analysis to identify novel genetic interactions potentially implicated in AD. We found that ERS can serve as an indicator of the genetic risk of AD.

Genome-wide pathogenesis interpretation using a heat diffusion-based systems genetics method and implications for gene function annotation.

BACKGROUND: Genetics is best dedicated to interpreting pathogenesis and revealing gene functions. The past decade has witnessed unprecedented progress in genetics, particularly in genome-wide identification of disorder variants through Genome-Wide Association Studies (GWAS) and Phenome-Wide Association Studies (PheWAS). However, it is still a great challenge to use GWAS/PheWAS-derived data to elucidate pathogenesis. METHODS: In this study, we used HotNet2, a heat diffusion-based systems genetics algorithm, to calculate the networks for disease genes obtained from GWAS and PheWAS, with an attempt to get deeper insights into disease pathogenesis at a molecular level. RESULTS: Through HotNet2 calculation, significant networks for 202 (for GWAS) and 167 (for PheWAS) types of diseases were identified and evaluated, respectively. The GWAS-derived disease networks exhibit a stronger biomedical relevance than PheWAS counterparts. Therefore, the GWAS-derived networks were used for pathogenesis interpretation by integrating the accumulated biomedical information. As a result, the pathogenesis for 64 diseases was elucidated in terms of mutation-caused abnormal transcriptional regulation, and 47 diseases were preliminarily interpreted in terms of mutation-caused varied protein-protein interactions. In addition, 3,802 genes (including 46 function-unknown genes) were assigned with new functions by disease network information, some of which were validated through mice gene knockout experiments. CONCLUSIONS: Systems genetics algorithm HotNet2 can efficiently establish genotype-phenotype links at the level of biological networks. Compared with original GWAS/PheWAS results, HotNet2-calculated disease-gene associations have stronger biomedical significance, hence provide better interpretations for the pathogenesis of genome-wide variants, and offer new insights into gene functions as well. These results are also helpful in drug development.