PRSet: Pathway-based polygenic risk score analyses and software

Schizophrenia polygenic risk scores, clinical variables and genetic pathways as predictors of phenotypic traits of bipolar I disorder

AIM

We investigated the predictive value of polygenic risk scores (PRS) derived from the schizophrenia GWAS (Trubetskoy et al., 2022) (SCZ3) for phenotypic traits of bipolar disorder type-I (BP-I) in 1878 BP-I cases and 2751 controls from Romania and UK.

METHODS

We used PRSice-v2.3.3 and PRS-CS for computing SCZ3-PRS for testing the predictive power of SCZ3-PRS alone and in combination with clinical variables for several BP-I subphenotypes and for pathway analysis. Non-linear predictive models were also used.

RESULTS

SCZ3-PRS significantly predicted psychosis, incongruent and congruent psychosis, general age-of-onset (AO) of BP-I, AO-depression, AO-Mania, rapid cycling in univariate regressions. A negative correlation between the number of depressive episodes and psychosis, mainly incongruent and an inverse relationship between increased SCZ3-SNP loading and BP-I-rapid cycling were observed. In random forest models comparing the predictive power of SCZ3-PRS alone and in combination with nine clinical variables, the best predictions were provided by combinations of SCZ3-PRS-CS and clinical variables closely followed by models containing only clinical variables. SCZ3-PRS performed worst. Twenty-two significant pathways underlying psychosis were identified.

LIMITATIONS

The combined RO-UK sample had a certain degree of heterogeneity of the BP-I severity: only the RO sample and partially the UK sample included hospitalized BP-I cases. The hospitalization is an indicator of illness severity. Not all UK subjects had complete subphenotype information.

CONCLUSION

Our study shows that the SCZ3-PRS have a modest clinical value for predicting phenotypic traits of BP-I. For clinical use their best performance is in combination with clinical variables.

Machine learning models for predicting blood pressure phenotypes by combining multiple polygenic risk scores

We construct non-linear machine learning (ML) prediction models for systolic and diastolic blood pressure (SBP, DBP) using demographic and clinical variables and polygenic risk scores (PRSs). We developed a two-model ensemble, consisting of a baseline model, where prediction is based on demographic and clinical variables only, and a genetic model, where we also include PRSs. We evaluate the use of a linear versus a non-linear model at both the baseline and the genetic model levels and assess the improvement in performance when incorporating multiple PRSs. We report the ensemble model's performance as percentage variance explained (PVE) on a held-out test dataset. A non-linear baseline model improved the PVEs from 28.1 to 30.1% (SBP) and 14.3% to 17.4% (DBP) compared with a linear baseline model. Including seven PRSs in the genetic model computed based on the largest available GWAS of SBP/DBP improved the genetic model PVE from 4.8 to 5.1% (SBP) and 4.7 to 5% (DBP) compared to using a single PRS. Adding additional 14 PRSs computed based on two independent GWASs further increased the genetic model PVE to 6.3% (SBP) and 5.7% (DBP). PVE differed across self-reported race/ethnicity groups, with primarily all non-White groups benefitting from the inclusion of additional PRSs. In summary, non-linear ML models improves BP prediction in models incorporating diverse populations.

Lack of genetic evidence for NLRP3-inflammasome involvement in Parkinson's disease pathogenesis

Activation of the NLRP3-inflammasome has been implicated in Parkinson's disease based on in vitro and in vivo studies. Clinical trials targeting the NLRP3-inflammasome in Parkinson's disease are ongoing. However, the evidence supporting NLRP3's involvement in Parkinson's disease from human genetics data is limited. In this study, we conducted analyses of common and rare variants in NLRP3-inflammasome related genes in Parkinson's disease cohorts. We performed pathway-specific analyses using polygenic risk scores and studied potential causal associations using Mendelian randomization with the NLRP3 components and the cytokines IL-1β and IL-18. Our findings showed no associations of common or rare variants, nor of the pathway polygenic risk score with Parkinson's disease. Mendelian randomization suggests that altering the expression of the NLRP3-inflammasome, IL-1β or IL-18, does not affect Parkinson's disease risk or progression. Therefore, our results do not support a role for the NLRP3-inflammasome in Parkinson's disease pathogenesis or as a target for drug development.

Dopamine signaling enriched striatal gene set predicts striatal dopamine synthesis and physiological activity in vivo

The polygenic architecture of schizophrenia implicates several molecular pathways involved in synaptic function. However, it is unclear how polygenic risk funnels through these pathways to translate into syndromic illness. Using tensor decomposition, we analyze gene co-expression in the caudate nucleus, hippocampus, and dorsolateral prefrontal cortex of post-mortem brain samples from 358 individuals. We identify a set of genes predominantly expressed in the caudate nucleus and associated with both clinical state and genetic risk for schizophrenia that shows dopaminergic selectivity. A higher polygenic risk score for schizophrenia parsed by this set of genes predicts greater dopamine synthesis in the striatum and greater striatal activation during reward anticipation. These results translate dopamine-linked genetic risk variation into in vivo neurochemical and hemodynamic phenotypes in the striatum that have long been implicated in the pathophysiology of schizophrenia.

Association of neurotransmitter pathway polygenic risk with specific symptom profiles in psychosis

A primary goal of psychiatry is to better understand the pathways that link genetic risk to psychiatric symptoms. Here, we tested association of diagnosis and endophenotypes with overall and neurotransmitter pathway-specific polygenic risk in patients with early-stage psychosis. Subjects included 205 demographically diverse cases with a psychotic disorder who underwent comprehensive psychiatric and neurological phenotyping and 115 matched controls. Following genotyping, we calculated polygenic scores (PGSs) for schizophrenia (SZ) and bipolar disorder (BP) using Psychiatric Genomics Consortium GWAS summary statistics. To test if overall genetic risk can be partitioned into affected neurotransmitter pathways, we calculated pathway PGSs (pPGSs) for SZ risk affecting each of four major neurotransmitter systems: glutamate, GABA, dopamine, and serotonin. Psychosis subjects had elevated SZ PGS versus controls; cases with SZ or BP diagnoses had stronger SZ or BP risk, respectively. There was no significant association within psychosis cases between individual symptom measures and overall PGS. However, neurotransmitter-specific pPGSs were moderately associated with specific endophenotypes; notably, glutamate was associated with SZ diagnosis and with deficits in cognitive control during task-based fMRI, while dopamine was associated with global functioning. Finally, unbiased endophenotype-driven clustering identified three diagnostically mixed case groups that separated on primary deficits of positive symptoms, negative symptoms, global functioning, and cognitive control. All clusters showed strong genome-wide risk. Cluster 2, characterized by deficits in cognitive control and negative symptoms, additionally showed specific risk concentrated in glutamatergic and GABAergic pathways. Due to the intensive characterization of our subjects, the present study was limited to a relatively small cohort. As such, results should be followed up with additional research at the population and mechanism level. Our study suggests pathway-based PGS analysis may be a powerful path forward to study genetic mechanisms driving psychiatric endophenotypes.

A machine learning approach for gene prioritization in Parkinson's disease

This scientific commentary refers to ‘Machine learning nominates the inositol pathway and novel genes in Parkinson’s disease’ by Yu et al. (https://doi.org/10.1093/brain/awad345).

Machine learning nominates the inositol pathway and novel genes in Parkinson's disease

There are 78 loci associated with Parkinson's disease in the most recent genome-wide association study (GWAS), yet the specific genes driving these associations are mostly unknown. Herein, we aimed to nominate the top candidate gene from each Parkinson's disease locus and identify variants and pathways potentially involved in Parkinson's disease. We trained a machine learning model to predict Parkinson's disease-associated genes from GWAS loci using genomic, transcriptomic and epigenomic data from brain tissues and dopaminergic neurons. We nominated candidate genes in each locus and identified novel pathways potentially involved in Parkinson's disease, such as the inositol phosphate biosynthetic pathway (INPP5F, IP6K2, ITPKB and PPIP5K2). Specific common coding variants in SPNS1 and MLX may be involved in Parkinson's disease, and burden tests of rare variants further support that CNIP3, LSM7, NUCKS1 and the polyol/inositol phosphate biosynthetic pathway are associated with the disease. Functional studies are needed to further analyse the involvements of these genes and pathways in Parkinson's disease.

Incorporating functional annotation with bilevel continuous shrinkage for polygenic risk prediction

BACKGROUND

Genetic variants can contribute differently to trait heritability by their functional categories, and recent studies have shown that incorporating functional annotation can improve the predictive performance of polygenic risk scores (PRSs). In addition, when only a small proportion of variants are causal variants, PRS methods that employ a Bayesian framework with shrinkage can account for such sparsity. It is possible that the annotation group level effect is also sparse. However, the number of PRS methods that incorporate both annotation information and shrinkage on effect sizes is limited. We propose a PRS method, PRSbils, which utilizes the functional annotation information with a bilevel continuous shrinkage prior to accommodate the varying genetic architectures both on the variant-specific level and on the functional annotation level.

RESULTS

We conducted simulation studies and investigated the predictive performance in settings with different genetic architectures. Results indicated that when there was a relatively large variability of group-wise heritability contribution, the gain in prediction performance from the proposed method was on average 8.0% higher AUC compared to the benchmark method PRS-CS. The proposed method also yielded higher predictive performance compared to PRS-CS in settings with different overlapping patterns of annotation groups and obtained on average 6.4% higher AUC. We applied PRSbils to binary and quantitative traits in three real world data sources (the UK Biobank, the Michigan Genomics Initiative (MGI), and the Korean Genome and Epidemiology Study (KoGES)), and two sources of annotations: ANNOVAR, and pathway information from the Kyoto Encyclopedia of Genes and Genomes (KEGG), and demonstrated that the proposed method holds the potential for improving predictive performance by incorporating functional annotations.

CONCLUSIONS

By utilizing a bilevel shrinkage framework, PRSbils enables the incorporation of both overlapping and non-overlapping annotations into PRS construction to improve the performance of genetic risk prediction. The software is available at https://github.com/styvon/PRSbils .

Evaluating the Efficacy of Type 2 Diabetes Polygenic Risk Scores in an Independent European Population

Numerous type 2 diabetes (T2D) polygenic risk scores (PGSs) have been developed to predict individuals' predisposition to the disease. An independent assessment and verification of the best-performing PGS are warranted to allow for a rapid application of developed models. To date, only 3% of T2D PGSs have been evaluated. In this study, we assessed all (n = 102) presently published T2D PGSs in an independent cohort of 3718 individuals, which has not been included in the construction or fine-tuning of any T2D PGS so far. We further chose the best-performing PGS, assessed its performance across major population principal component analysis (PCA) clusters, and compared it with newly developed population-specific T2D PGS. Our findings revealed that 88% of the published PGSs were significantly associated with T2D; however, their performance was lower than what had been previously reported. We found a positive association of PGS improvement over the years (p-value = 8.01 × 10-4 with PGS002771 currently showing the best discriminatory power (area under the receiver operating characteristic (AUROC) = 0.669) and PGS003443 exhibiting the strongest association PGS003443 (odds ratio (OR) = 1.899). Further investigation revealed no difference in PGS performance across major population PCA clusters and when compared with newly developed population-specific PGS. Our findings revealed a positive trend in T2D PGS performance, consistently identifying high-T2D-risk individuals in an independent European population.

Unraveling the intercellular communication disruption and key pathways in Alzheimer's disease: an integrative study of single-nucleus transcriptomes and genetic association

BACKGROUND

Recently, single-nucleus RNA-seq (snRNA-seq) analyses have revealed important cellular and functional features of Alzheimer's disease (AD), a prevalent neurodegenerative disease. However, our knowledge regarding intercellular communication mediated by dysregulated ligand-receptor (LR) interactions remains very limited in AD brains.

METHODS

We systematically assessed the intercellular communication networks by using a discovery snRNA-seq dataset comprising 69,499 nuclei from 48 human postmortem prefrontal cortex (PFC) samples. We replicated the findings using an independent snRNA-seq dataset of 56,440 nuclei from 18 PFC samples. By integrating genetic signals from AD genome-wide association studies (GWAS) summary statistics and whole genome sequencing (WGS) data, we prioritized AD-associated Gene Ontology (GO) terms containing dysregulated LR interactions. We further explored drug repurposing for the prioritized LR pairs using the Therapeutic Targets Database.

RESULTS

We identified 190 dysregulated LR interactions across six major cell types in AD PFC, of which 107 pairs were replicated. Among the replicated LR signals, we found globally downregulated communications in the astrocytes-to-neurons signaling axis, characterized, for instance, by the downregulation of APOE-related and Calmodulin (CALM)-related LR interactions and their potential regulatory connections to target genes. Pathway analyses revealed 44 GO terms significantly linked to AD, highlighting Biological Processes such as 'amyloid precursor protein processing' and 'ion transmembrane transport,' among others. We prioritized several drug repurposing candidates, such as cromoglicate, targeting the identified dysregulated LR pairs.

CONCLUSIONS

Our integrative analysis identified key dysregulated LR interactions in a cell type-specific manner and the associated GO terms in AD, offering novel insights into potential therapeutic targets involved in disrupted cell-cell communication in AD.

Genetic landscape of Parkinson's disease and related diseases in Luxembourg

Objectives

To explore the genetic architecture of PD in the Luxembourg Parkinson's Study including cohorts of healthy people and patients with Parkinson's disease (PD) and atypical parkinsonism (AP).

Methods

809 healthy controls, 680 PD and 103 AP were genotyped using the Neurochip array. We screened and validated rare single nucleotide variants (SNVs) and copy number variants (CNVs) within seven PD-causing genes (LRRK2, SNCA, VPS35, PRKN, PARK7, PINK1 and ATP13A2). Polygenic risk scores (PRSs) were generated using the latest genome-wide association study for PD. We then estimated the role of common variants in PD risk by applying gene-set-specific PRSs.

Results

We identified 60 rare SNVs in seven PD-causing genes, nine of which were pathogenic in LRRK2, PINK1 and PRKN. Eleven rare CNVs were detected in PRKN including seven duplications and four deletions. The majority of PRKN SNVs and CNVs carriers were heterozygous and not differentially distributed between cases and controls. The PRSs were significantly associated with PD and identified specific molecular pathways related to protein metabolism and signal transduction as drivers of PD risk.

Conclusion

We performed a comprehensive genetic characterization of the deep-phenotyped individuals of the Luxembourgish Parkinson's Study. Heterozygous SNVs and CNVs in PRKN were not associated with higher PD risk. In particular, we reported novel digenic variants in PD related genes and rare LRRK2 SNVs in AP patients. Our findings will help future studies to unravel the genetic complexity of PD.

Machine learning models for blood pressure phenotypes combining multiple polygenic risk scores

We construct non-linear machine learning (ML) prediction models for systolic and diastolic blood pressure (SBP, DBP) using demographic and clinical variables and polygenic risk scores (PRSs). We developed a two-model ensemble, consisting of a baseline model, where prediction is based on demographic and clinical variables only, and a genetic model, where we also include PRSs. We evaluate the use of a linear versus a non-linear model at both the baseline and the genetic model levels and assess the improvement in performance when incorporating multiple PRSs. We report the ensemble model's performance as percentage variance explained (PVE) on a held-out test dataset. A non-linear baseline model improved the PVEs from 28.1% to 30.1% (SBP) and 14.3% to 17.4% (DBP) compared with a linear baseline model. Including seven PRSs in the genetic model computed based on the largest available GWAS of SBP/DBP improved the genetic model PVE from 4.8% to 5.1% (SBP) and 4.7% to 5% (DBP) compared to using a single PRS. Adding additional 14 PRSs computed based on two independent GWASs further increased the genetic model PVE to 6.3% (SBP) and 5.7% (DBP). PVE differed across self-reported race/ethnicity groups, with primarily all non-White groups benefitting from the inclusion of additional PRSs.

Cell-type-specific Alzheimer's disease polygenic risk scores are associated with distinct disease processes in Alzheimer's disease

Many of the Alzheimer's disease (AD) risk genes are specifically expressed in microglia and astrocytes, but how and when the genetic risk localizing to these cell types contributes to AD pathophysiology remains unclear. Here, we derive cell-type-specific AD polygenic risk scores (ADPRS) from two extensively characterized datasets and uncover the impact of cell-type-specific genetic risk on AD endophenotypes. In an autopsy dataset spanning all stages of AD (n = 1457), the astrocytic ADPRS affected diffuse and neuritic plaques (amyloid-β), while microglial ADPRS affected neuritic plaques, microglial activation, neurofibrillary tangles (tau), and cognitive decline. In an independent neuroimaging dataset of cognitively unimpaired elderly (n = 2921), astrocytic ADPRS was associated with amyloid-β, and microglial ADPRS was associated with amyloid-β and tau, connecting cell-type-specific genetic risk with AD pathology even before symptom onset. Together, our study provides human genetic evidence implicating multiple glial cell types in AD pathophysiology, starting from the preclinical stage.

Psychosis Endophenotypes: A Gene-Set-Specific Polygenic Risk Score Analysis

BACKGROUND AND HYPOTHESIS

Endophenotypes can help to bridge the gap between psychosis and its genetic predispositions, but their underlying mechanisms remain largely unknown. This study aims to identify biological mechanisms that are relevant to the endophenotypes for psychosis, by partitioning polygenic risk scores into specific gene sets and testing their associations with endophenotypes.

STUDY DESIGN

We computed polygenic risk scores for schizophrenia and bipolar disorder restricted to brain-related gene sets retrieved from public databases and previous publications. Three hundred and seventy-eight gene-set-specific polygenic risk scores were generated for 4506 participants. Seven endophenotypes were also measured in the sample. Linear mixed-effects models were fitted to test associations between each endophenotype and each gene-set-specific polygenic risk score.

STUDY RESULTS

After correction for multiple testing, we found that a reduced P300 amplitude was associated with a higher schizophrenia polygenic risk score of the forebrain regionalization gene set (mean difference per SD increase in the polygenic risk score: -1.15 µV; 95% CI: -1.70 to -0.59 µV; P = 6 × 10-5). The schizophrenia polygenic risk score of forebrain regionalization also explained more variance of the P300 amplitude (R2 = 0.032) than other polygenic risk scores, including the genome-wide polygenic risk scores.

CONCLUSIONS

Our finding on reduced P300 amplitudes suggests that certain genetic variants alter early brain development thereby increasing schizophrenia risk years later. Gene-set-specific polygenic risk scores are a useful tool to elucidate biological mechanisms of psychosis and endophenotypes, offering leads for experimental validation in cellular and animal models.

Treatment resistance NMDA receptor pathway polygenic score is associated with brain glutamate in schizophrenia

Dysfunction of glutamate neurotransmission has been implicated in the pathophysiology of schizophrenia and may be particularly relevant in severe, treatment-resistant symptoms. The underlying mechanism may involve hypofunction of the NMDA receptor. We investigated whether schizophrenia-related pathway polygenic scores, composed of genetic variants within NMDA receptor encoding genes, are associated with cortical glutamate in schizophrenia. Anterior cingulate cortex (ACC) glutamate was measured in 70 participants across 4 research sites using Proton Magnetic Resonance Spectroscopy (1H-MRS). Two NMDA receptor gene sets were sourced from the Molecular Signatories Database and NMDA receptor pathway polygenic scores were constructed using PRSet. The NMDA receptor pathway polygenic scores were weighted by single nucleotide polymorphism (SNP) associations with treatment-resistant schizophrenia, and associations with ACC glutamate were tested. We then tested whether NMDA receptor pathway polygenic scores with SNPs weighted by associations with non-treatment-resistant schizophrenia were associated with ACC glutamate. A higher NMDA receptor complex pathway polygenic score was significantly associated with lower ACC glutamate (β = -0.25, 95 % CI = -0.49, -0.02, competitive p = 0.03). When SNPs were weighted by associations with non-treatment-resistant schizophrenia, there was no association between the NMDA receptor complex pathway polygenic score and ACC glutamate (β = 0.05, 95 % CI = -0.18, 0.27, competitive p = 0.79). These results provide initial evidence of an association between common genetic variation implicated in NMDA receptor function and ACC glutamate levels in schizophrenia. This association was specific to when the NMDA receptor complex pathway polygenic score was weighted by SNP associations with treatment-resistant schizophrenia.

Dopamine and schizophrenia from bench to bedside: Discovery of a striatal co-expression risk gene set that predicts in vivo measures of striatal function

Schizophrenia (SCZ) is characterized by a polygenic risk architecture implicating diverse molecular pathways important for synaptic function. However, how polygenic risk funnels through these pathways to translate into syndromic illness is unanswered. To evaluate biologically meaningful pathways of risk, we used tensor decomposition to characterize gene co-expression in post-mortem brain (of neurotypicals: N=154; patients with SCZ: N=84; and GTEX samples N=120) from caudate nucleus (CN), hippocampus (HP), and dorsolateral prefrontal cortex (DLPFC). We identified a CN-predominant gene set showing dopaminergic selectivity that was enriched for genes associated with clinical state and for genes associated with SCZ risk. Parsing polygenic risk score for SCZ based on this specific gene set (parsed-PRS), we found that greater pathway-specific SCZ risk predicted greater in vivo striatal dopamine synthesis capacity measured by [ 18 F]-FDOPA PET in three independent cohorts of neurotypicals and patients (total N=235) and greater fMRI striatal activation during reward anticipation in two additional independent neurotypical cohorts (total N=141). These results reveal a 'bench to bedside' translation of dopamine-linked genetic risk variation in driving in vivo striatal neurochemical and hemodynamic phenotypes that have long been implicated in the pathophysiology of SCZ.

Pathway-Specific Polygenic Scores Improve Cross-Ancestry Prediction of Psychosis and Clinical Outcomes

Psychotic disorders are debilitating conditions with disproportionately high public health burden. Genetic studies indicate high heritability, but current polygenic scores (PGS) account for only a fraction of variance in psychosis risk. PGS often show poor portability across ancestries, performing significantly worse in non-European populations. Pathway-specific PGS (pPGS), which restrict PGS to genomic locations within distinct biological units, could lead to increased mechanistic understanding of pathways that lead to risk and improve cross-ancestry prediction by reducing noise in genetic predictors. This study examined the predictive power of genome-wide PGS and nine pathway-specific pPGS in a unique Chinese-ancestry sample of deeply-phenotyped psychosis patients and non-psychiatric controls. We found strong evidence for the involvement of schizophrenia-associated risk variants within "nervous system development" (p=2.5e-4) and "regulation of neuron differentiation" pathways (p=3.0e-4) in predicting risk for psychosis. We also found the "ion channel complex" pPGS, with weights derived from GWAS of bipolar disorder, to be strongly associated with the number of inpatient psychiatry admissions a patient experiences (p=1.5e-3) and account for a majority of the signal in the overall bipolar PGS. Importantly, although the schizophrenia genome-wide PGS alone explained only 3.7% of the variance in liability to psychosis in this Chinese ancestry sample, the addition of the schizophrenia-weighted pPGS for "nervous system development" and "regulation of neuron differentiation" increased the variance explained to 6.9%, which is on-par with the predictive power of PGS in European ancestry samples. Thus, not only can pPGS provide greater insight into mechanisms underlying genetic risk for disease and clinical outcomes, but may also improve cross-ancestry risk prediction accuracy.

Unraveling the intercellular communication disruption and key pathways in Alzheimer's disease: An integrative study of single-nucleus transcriptomes and genetic association

Background

Recently, single-nucleus RNA-seq (snRNA-seq) analyses have revealed important cellular and functional features of Alzheimer's disease (AD), a prevalent neurodegenerative disease. However, our knowledge regarding intercellular communication mediated by dysregulated ligand-receptor (LR) interactions remains very limited in AD brains.

Methods

We systematically assessed the intercellular communication networks by using a discovery snRNA-seq dataset comprising 69,499 nuclei from 48 human postmortem prefrontal cortex (PFC) samples. We replicated the findings using an independent snRNA-seq dataset of 56,440 nuclei from 18 PFC samples. By integrating genetic signals from AD genome-wide association studies (GWAS) summary statistics and whole genome sequencing (WGS) data, we prioritized AD-associated Gene Ontology (GO) terms containing dysregulated LR interactions. We further explored drug repurposing for the prioritized LR pairs using the Therapeutic Targets Database.

Results

We identified 316 dysregulated LR interactions across six major cell types in AD PFC, of which 210 pairs were replicated. Among the replicated LR signals, we found globally downregulated communications in astrocytes-to-neurons signaling axis, characterized, for instance, by the downregulation of APOE-related and Calmodulin (CALM)-related LR interactions and their potential regulatory connections to target genes. Pathway analyses revealed 60 GO terms significantly linked to AD, highlighting Biological Processes such as 'amyloid precursor protein processing' and 'ion transmembrane transport', among others. We prioritized several drug repurposing candidates, such as cromoglicate, targeting the identified dysregulated LR pairs.

Conclusions

Our integrative analysis identified key dysregulated LR interactions in a cell type-specific manner and the associated GO terms in AD, offering novel insights into potential therapeutic targets involved in disrupted cell-cell communication in AD.

Minimizing enzyme mass to decompose flux distribution for identifying biologically relevant elementary flux modes

The flux distribution in metabolic network can be decomposed as non-negative linear combinations of elementary flux modes (EFMs). Identifying biologically relevant EFM combination by decomposing flux distribution in metabolic network is a useful method to study metabolisms in systems biology. However, the occurrence of biologically irrelevant EFMs hinders the application of such methods. In this paper, we introduce a novel method for identifying EFM combination by minimizing enzyme mass. Our proposed method, called EMMD (Enzyme Mass Minimization Decomposition), takes into consideration both thermodynamic and enzymatic constraints in stoichiometry metabolic models. By implementing EMMD, we can decompose the flux distributions in metabolic network to detect biologically relevant EFM combinations. We demonstrate the effectiveness of our method by applying it to the core Escherichia coli metabolic network and show that the optimal EFM combinations identified by EMMD are unique. Moreover, the optimal EFM combination identified by EMMD not only aligns more closely with experimental values in terms of estimated growth rate, but it also demonstrates more favorable thermodynamics. Finally, we investigated the growth of the core Escherichia coli metabolic network in Luria-Bertani medium containing different carbon sources, revealing the impact of various carbon sources on the growth rate of flux distribution. EMMD thus could be a promising complement to the existing flux decomposition tools.

Genetically-regulated pathway-polygenic risk score (GRPa-PRS): A risk stratification method to identify genetically regulated pathways in polygenic diseases

Background

Alzheimer's disease (AD) is a common neurodegenerative disease in the elderly population, with genetic factors playing an important role. A considerable proportion of elderly people carry a high genetic AD risk but evade AD. On the other hand, some individuals with a low risk for AD eventually develop AD. We hypothesized that unknown counterfactors might be involved in reversing the polygenic risk scores (PRS) prediction, which might provide insights into AD pathogenesis, prevention, and early clinical intervention.

Methods

We built a novel computational framework to identify genetically-regulated pathways (GRPa) using PRS-based stratification for each cohort. We curated two AD cohorts with genotyping data; the discovery and the replication dataset include 2722 and 2492 individuals, respectively. First, we calculated the optimized PRS model based on the three latest AD GWAS summary statistics for each cohort. Then, we sub-grouped the individuals by their PRS and clinical diagnosis into groups such as cognitively normal (CN) with high PRS for AD (resilient group), AD cases with low PRS (susceptible group), and AD/CNs participants with similar PRS backgrounds. Lastly, we imputed the individual genetically-regulated expression (GReX) and identified the differential GRPas between subgroups with gene-set enrichment analysis and gene-set variational analysis in 2 models with and without the effect of APOE.

Results

For each subgroup, we conducted the same procedures in both the discovery and replication datasets across three PRS models for comparison. In Model 1 with the APOE region, we identified well-known AD-related pathways, including amyloid-beta clearance, tau protein binding, and astrocytes response to oxidative stress. In Model 2 without the APOE region, synapse function, microglia function, histidine metabolism, and thiolester hydrolase activity were significant, suggesting that they are pathways independent of the effect of APOE. Finally, our GRPa-PRS method reduces the false discovery rate in detecting differential pathways compared to another variants-based pathway PRS method.

Conclusions

We developed a framework, GRPa-PRS, to systematically explore the differential GRPas among individuals stratified by their estimated PRS. The GReX-level comparison among those groups unveiled new insights into the pathways associated with AD risk and resilience. Our framework can be extended to other polygenic complex diseases.

Cell-type-specific Alzheimer's disease polygenic risk scores are associated with distinct disease processes in Alzheimer's disease

Alzheimer's disease (AD) heritability is enriched in glial genes, but how and when cell-type-specific genetic risk contributes to AD remains unclear. Here, we derive cell-type-specific AD polygenic risk scores (ADPRS) from two extensively characterized datasets. In an autopsy dataset spanning all stages of AD (n=1,457), astrocytic (Ast) ADPRS was associated with both diffuse and neuritic Aβ plaques, while microglial (Mic) ADPRS was associated with neuritic Aβ plaques, microglial activation, tau, and cognitive decline. Causal modeling analyses further clarified these relationships. In an independent neuroimaging dataset of cognitively unimpaired elderly (n=2,921), Ast-ADPRS were associated with Aβ, and Mic-ADPRS was associated with Aβ and tau, showing a consistent pattern with the autopsy dataset. Oligodendrocytic and excitatory neuronal ADPRSs were associated with tau, but only in the autopsy dataset including symptomatic AD cases. Together, our study provides human genetic evidence implicating multiple glial cell types in AD pathophysiology, starting from the preclinical stage.

Genetic Determinants of the Acute Respiratory Distress Syndrome

Acute respiratory distress syndrome (ARDS) is a life-threatening lung condition that arises from multiple causes, including sepsis, pneumonia, trauma, and severe coronavirus disease 2019 (COVID-19). Given the heterogeneity of causes and the lack of specific therapeutic options, it is crucial to understand the genetic and molecular mechanisms that underlie this condition. The identification of genetic risks and pharmacogenetic loci, which are involved in determining drug responses, could help enhance early patient diagnosis, assist in risk stratification of patients, and reveal novel targets for pharmacological interventions, including possibilities for drug repositioning. Here, we highlight the basis and importance of the most common genetic approaches to understanding the pathogenesis of ARDS and its critical triggers. We summarize the findings of screening common genetic variation via genome-wide association studies and analyses based on other approaches, such as polygenic risk scores, multi-trait analyses, or Mendelian randomization studies. We also provide an overview of results from rare genetic variation studies using Next-Generation Sequencing techniques and their links with inborn errors of immunity. Lastly, we discuss the genetic overlap between severe COVID-19 and ARDS by other causes.

Distinct genetic liability profiles define clinically relevant patient strata across common diseases

Genome-wide association studies have unearthed a wealth of genetic associations across many complex diseases. However, translating these associations into biological mechanisms contributing to disease etiology and heterogeneity has been challenging. Here, we hypothesize that the effects of disease-associated genetic variants converge onto distinct cell type specific molecular pathways within distinct subgroups of patients. In order to test this hypothesis, we develop the CASTom-iGEx pipeline to operationalize individual level genotype data to interpret personal polygenic risk and identify the genetic basis of clinical heterogeneity. The paradigmatic application of this approach to coronary artery disease and schizophrenia reveals a convergence of disease associated variant effects onto known and novel genes, pathways, and biological processes. The biological process specific genetic liabilities are not equally distributed across patients. Instead, they defined genetically distinct groups of patients, characterized by different profiles across pathways, endophenotypes, and disease severity. These results provide further evidence for a genetic contribution to clinical heterogeneity and point to the existence of partially distinct pathomechanisms across patient subgroups. Thus, the universally applicable approach presented here has the potential to constitute an important component of future personalized medicine concepts.

Dissecting the polygenic basis of atherosclerosis via disease-associated cell state signatures

Coronary artery disease (CAD) is a pandemic disease where up to half of the risk is explained by genetic factors. Advanced insights into the genetic basis of CAD require deeper understanding of the contributions of different cell types, molecular pathways, and genes to disease heritability. Here, we investigate the biological diversity of atherosclerosis-associated cell states and interrogate their contribution to the genetic risk of CAD by using single-cell and bulk RNA sequencing (RNA-seq) of mouse and human lesions. We identified 12 disease-associated cell states that we characterized further by gene set functional profiling, ligand-receptor prediction, and transcription factor inference. Importantly, Vcam1+ smooth muscle cell state genes contributed most to SNP-based heritability of CAD. In line with this, genetic variants near smooth muscle cell state genes and regulatory elements explained the largest fraction of CAD-risk variance between individuals. Using this information for variant prioritization, we derived a hybrid polygenic risk score (PRS) that demonstrated improved performance over a classical PRS. Our results provide insights into the biological mechanisms associated with CAD risk, which could make a promising contribution to precision medicine and tailored therapeutic interventions in the future.

Non-additive effects of schizophrenia risk genes reflect convergent downstream function

Genetic studies of schizophrenia (SCZ) reveal a complex polygenic risk architecture comprised of hundreds of risk variants, the majority of which are common in the population at-large and confer only modest increases in disorder risk. Precisely how genetic variants with individually small predicted effects on gene expression combine to yield substantial clinical impacts in aggregate is unclear. Towards this, we previously reported that the combinatorial perturbation of four SCZ risk genes ("eGenes", whose expression is regulated by common variants) resulted in gene expression changes that were not predicted by individual perturbations, being most non-additive among genes associated with synaptic function and SCZ risk. Now, across fifteen SCZ eGenes, we demonstrate that non-additive effects are greatest within groups of functionally similar eGenes. Individual eGene perturbations reveal common downstream transcriptomic effects ("convergence"), while combinatorial eGene perturbations result in changes that are smaller than predicted by summing individual eGene effects ("sub-additive effects"). Unexpectedly, these convergent and sub-additive downstream transcriptomic effects overlap and constitute a large proportion of the genome-wide polygenic risk score, suggesting that functional redundancy of eGenes may be a major mechanism underlying non-additivity. Single eGene perturbations likewise fail to predict the magnitude or directionality of cellular phenotypes resulting from combinatorial perturbations. Overall, our results indicate that polygenic risk cannot be extrapolated from experiments testing one risk gene at a time and must instead be empirically measured. By unravelling the interactions between complex risk variants, it may be possible to improve the clinical utility of polygenic risk scores through more powerful prediction of symptom onset, clinical trajectory, and treatment response, or to identify novel targets for therapeutic intervention.