1. Can we move from genetics to pathways to (new) drugs?
Dr. Gerome Breen, PGC Network and Pathways Group, NIHR BRC for Mental Health, King's College London
Meeting introduction and general aims.
2. Discover More; Illumina portfolio update
Dr. Cora Vacher, Illumina
As genomics moves to bigger scale, and deeper biological understanding, the need for scalable, sensitive tools has also increased. We present the latest solutions from Illumina for characterising gene expression, epigenomics, and drug development.
3. A Pharma Perspective on Drug Development from Genomics
Dr. David Collier, Eli Lilly and Company
By making direct disease associations, human genetics can improve the probability of identifying a valid drug target, i.e. a target that when modulated pharmacologically, provides meaningful efficacy and acceptable safety for specific human disease in long- term clinical usage. The path to validation requires a specific hypothesis about the desired pharmacological action resulting from a compound binding to the molecular entity (e.g. agonist of or inhibition of some biological effect) that will halt/slow/reverse the disease, and a sufficient body of scientific evidence to persuade the scientific/medical/ commercial community to invest in developing the "binder" in to test the hypothesis specified in patients. Validation is subjective and dynamic, situational and specific: every drug target is different. The validation process is also never perfect: there is a hypothesis to test in patients. Given the high cost of drug development and the range of unmet need across many types of disease and disorder, particular indications have to have a level of tractability to make them realistic propositions for drug development. In the absence of this (for example lack of knowledge of disease pathophysiology), drug discovery may stall. The present talk will discuss some of the steps that may be needed to restart this process in disease areas which have become less attractive propositions.
4. A Pharmademic View of Drug Discovery
Dr. Michael R. Barnes, Genomics England
Arguably the momentum of drug discovery as an exclusively industrial activity has stalled. Opinion on the underlying causes of this slow down are divided, some suggesting that the low hanging fruit among new medicines have mostly been found, whilst others point to a more generalised failure to translate from target to the clinic. This lowered productivity is set against a backdrop of unprecedented public investment in the life sciences, particularly focused on fostering industry-academic partnerships, exemplified in the EU by the Innovative Medicines Initiative, Open Targets and the UK 100K genome project and in the US by the Precision Medicine initiative. The widespread engagement of pharma in these initiatives, acknowledges a shift in internal R&D focus towards the later stages of development and an increasing reliance on academic partnership for early stage drug discovery and clinical translation. At the same time, the rewards in drug discovery are increasingly being reaped by those with the tools to leverage biomedical “big data”. Arguably, computational methods now are equal in translational importance to the laboratory. The concept of computational drug repositioning is already well established, but now computational methods are also offering insight into drug mode of action, stratified medicine and target validation. These and other opportunities in computational biology and chemistry are illustrated, along with challenges that still remain between bench and bedside.
5. Janssen Psychiatric Genetic Assets, Research Interest and Opportunities
Dr. Qingqin Li, Johnson & Johnson
In this talk, I will summarize the Janssen psychiatric genetic assets and past/present endeavors, highlight the current research focus within Mood Disease Area Stronghold (DAS). I will also discuss potential opportunities in pre-competitive space for academy and pharma to work together and explore ways to maximize the value of Janssen samples and data.
6. Translational Model Development for Psychiatry Research
Dr. Niels Plath, Lundbeck
The etiological and phenotypical heterogeneity of psychiatry disorders such as schizophrenia precludes the design of disease models in animal species. Increasing insight into intermediate phenotypes of schizophrenia, however, provides an opportunity to address conserved biological mechanisms with translational readouts across species. Building the subsequent link of such well-described, distinct biological phenomena to clinical symptoms is a huge but necessary task. Rare genetic risk factors with high penetrance to schizophrenia, here the copy number variant 15q13.3, are linked to an inability to generate brain oscillations in a mouse model organism. Such dysfunction can be assessed translational across species and thereby linked to symptomatic features. This presentation provides an example of an endophenotype linked to schizophrenia and the attempt to utilize this as a target for future pharmacological intervention.
7. The Psychiatric Genomics Consortium: Finding Actionable Variation
Prof. Patrick Sullivan, UNC Chapel Hill & Karolinska Institutet
8. Anorexia Reconceptualized: Genetic Evidence for Psychiatric and Metabolic Etiology
Dr. Cynthia Bulik, UNC Chapel Hill & Karolinska Institutet
The Eating Disorders Workgroup of the Psychiatric Genomics Consortium presents the first genome-wide significant locus for anorexia nervosa along with genetic correlations which reveal intriguing associations with a range of psychiatric and metabolic parameters. In addition, we present longitudinal prospective evidence for negative deviations from BMI-based growth trajectories as early as age 2 in females and age 7 in males in individuals who later develop anorexia nervosa. Parallel positive growth curve deviations emerge for those who go on to develop binge-eating disorder. Our observations encourage a reconceptualization of eating disorders as having both psychiatric and metabolic underpinnings. This may represent an instance in which a rare disorder can inform a global public health concern, namely the obesity epidemic.
9. First Wave Genome Wide Study in PTSD: Genetic Overlap and Sex Differences in Heritability
Dr. Israel Liberzon, University of Michigan
The Psychiatric Genomics Consortium PTSD group combined genome-wide case-control molecular genetic data across 11 multi-ethnic studies (19 GWAS - tiered meta analyses) to quantify PTSD heritability, to examine shared genetic risk with schizophrenia, bipolar, and major depressive disorder, and to identify risk loci for PTSD. In the first wave analysis we examined 20 730 individuals, and found a molecular genetics-based heritability estimate (h2SNP) for European American females of 29%, which is similar to h2SNP for schizophrenia and is substantially higher than h2SNP in European American males (estimate not distinguishable from zero). There was also strong evidence of overlapping genetic risk between PTSD and schizophrenia and more modest evidence of overlap with bipolar and major depressive disorder. No SNPs exceeded genome wide significance in the overall meta-analysis, nor did we replicate previously reported associations from individual cohorts. However, SNP-level summary statistics afford the best-available molecular genetic index of PTSD and can be used in polygenic risk prediction and genetic correlation studies . Furthermore, our unique data contributes to the broader goal of increased ancestral diversity in genomic data resources. In sum, results demonstrate marked heritability for the development of PTSD in women, shared genetic risk between PTSD and other psychiatric disorders, and highlight the importance of multi-ethnic/racial samples.
10. PGC3 Large Pedigree Sequencing Project
Dr. Michael Gill, Trinity College Dublin
Direct selection influences the genetic architectures so that common variants have very small effects (GRR<1.2). Rare mutations are a sizable reservoir of actionable genetic variation that could be missed by SNP arrays. De novo mutations can have large effects on ASD and SCZ (GRR 5-20). Mutations can segregate for a few generations leading to densely affected pedigrees.
We will select multiply affected pedigrees for WGS. We will derive GRS using the SNP array. From GRS, we will estimate the probability of the observed pattern of inheritance. For eg. if a pedigree mean GRS is at the population mean (lifetime risk ~1%), observing 5 of 10 affected is notable (P=2x10-8).
Variants will be prioritized and incorporate frequency data from ExAC, and co-segregation with phenotype. Any putative causal mutation requires: (a) statistically compelling co-segregation in the pedigree, (b) genetic replication in external studies (independent pedigrees, see below, or in case-control samples), and (c) supported by functional evidence.
We will combine families with positive LOD scores in a region. If LOD scores reflect chance, they will not be underpinned by rare variants in the same gene in different pedigrees. Second, WGS allows a genotype-first approach to linkage across families. For each gene, we consider only families where one affected carries a rare “prioritized” variant.
11. UK Biobank Mental Health Phenotyping
Dr. Matthew Hotopf , Director, NIHR BRC for Mental Health, King's College London
12. Drug Discovery: from Academe to Industry
Dr. Allan Young, King's College London
Common psychiatric conditions such as anxiety, mood and psychotic disorders are the leading cause of ill health in the world today. Other disease areas, such as cardiovascular and cancer, are benefiting from novel therapeutics which are the result of preceding advances in basic science leading to improved understanding of disease mechanisms and resulting drug discovery. Such advances have been delayed in psychiatry due to relative under-investment in research and the inherent complexity of the brain and behaviour. These factors notwithstanding, advances in genetics, biomarkers, neuroimaging and related areas of basic and clinical neuroscience set the scene for future evidence based drug discovery in psychiatry.
13. Challenges in Drug Development – A Data-driven Analysis
Dr. John Overington, BenevolentAI
The drug discovery industry is facing major systematic failure in discovering novel drugs. Although some of this lack of productivity may be attributed to the low hanging fruit already being harvested, leaving more challenging diseases and interventions to develop, there are other factors at play. A key recent change in technology has been the essentially free and ready access to low cost computing and data; however, it is clear that this alone is not enough, and smart integration and hypothesis generation and in silico validation across this data will be essential before impacts on progress are realised. In this presentation I discuss the building of a large public drug discovery database connecting compounds through to pharmacological effects and molecular targets, then provide an overview and comparison of successful and unsuccessful drug discovery and development programs, leading into insights for future likely successful target systems, then this view is integrated into the framework of large-scale personalised medicine. Finally, the issue of scientific reproducibility, and it’s impact on data analysis is discussed.
14. Analysis and Visualisation of Drug Genetic Pathways
Dr. Héléna A. Gaspar, PGC, King's College London
Harnessing results from genome-wide association studies to find new potential treatments for psychiatric disorders is an on-going challenge. For schizophrenia, larger sample sizes in GWAS studies resulted in higher enrichment for antipsychotics and epileptics, indicating that GWAS results could be used for drug repurposing. Growing databases and increasing sample sizes (cf. Psychiatric Genomics Consortium GWASs) should allow us to highlight new drug repurposing opportunities in the near future. Drug/target and drug/gene interactions mined from public databases (ChEMBL, DGIdb, Ki DB, PHAROS...) can be used to generate drug gene-sets, and online ontologies and databases such as Open Targets, GO, and MSigDB, can be used to construct disease pathways and biological pathways. The association between gene-sets and psychiatric disorders is often assessed with pathway analysis tools, but further interpretation of pathways is often lacking. This talks illustrates how the overlap between pathways in "gene space" and the distribution of genetic association within a pathway can be visualised using dimensionality reduction methods. Pathways should be investigated as subnetworks: their association with a specific disorder cannot be reduced to a single p-value or other association statistic, and visualisation is essential.
15. Illuminating the Druggable GPCR-ome
Dr. Bryan Roth, UNC Chapel Hill
I will discuss mainly public efforts to illuminate under-studied targets of potential relevance for neuropsychiatric diseases as illuminated by the PGC effort. I will focus on a recent initiative 'Illuminating the Druggable Genome' and highlight open-source (PHAROS; https://pharos.nih.gov/idg/index) handled by the Druggable Genome Knowledge Management Center (IDG KMC) which evaluates, organizes and distills more than 80 protein-centric and 20 gene-centric resources for over 20,000 curated human proteins as well as resources from the NIMH Psychoactive Drug Screening Program (http://pdspdb.unc.edu/pdspWeb/).
16. Using A.I. to Solve the Complexity of Designing Psychiatric Drugs
Dr. Andrew Hopkins, University of Dundee, Exscientia Ltd.
Our deepening understanding of the genomics of psychiatric diseases holds the promise of better translational to the clinic. However, the price of our deeper understanding is an increase in the complexity of the drug discovery process for new psychiatric drugs. The polygenic model of psychiatric diseases suggests there maybe few stand-alone, individual drug targets. Therefore, we need to consider complex ‘target product profiles’ for new medicines that deliberately work by synergy and emergent behavioural properties, by acting on multiple drug targets in parallel. Indeed, historically, most treatment options we have today for psychiatric illnesses act on multiple CNS drug targets, via polypharmacology. Until now successful, multi-target CNS drugs were not designed rationally, but were often discovered by phenotypic screening or clinical observations. To tackle the complexity in psychiatric drug discovery, Exscientia Ltd has developed an Artificial Intelligence (A.I.) platform to design drugs rationally towards desired polypharmacology profiles or phenotypic end-points in high content assays. We shall describe in this presentation our on-going work on the development of the next generation of psychiatric drugs, that are being designed by algorithm.
17. iPsych and PGC ADHD and Autism Workgroups Update
Dr. Anders Børglum, PGC iPsych, Aarhus University
The Psychiatric Genomics Consortium PTSD group combined genome-wide case-control molecular genetic data across 11 multi-ethnic studies (19 GWAS - tiered meta analyses) to quantify PTSD heritability, to examine shared genetic risk with schizophrenia, bipolar, and major depressive disorder, and to identify risk loci for PTSD. In the first wave analysis we examined 20 730 individuals, and found a molecular genetics-based heritability estimate (h2SNP) for European American females of 29%, which is similar to h2SNP for schizophrenia and is substantially higher than h2SNP in European American males (estimate not distinguishable from zero). There was also strong evidence of overlapping genetic risk between PTSD and schizophrenia and more modest evidence of overlap with bipolar and major depressive disorder. No SNPs exceeded genome wide significance in the overall meta-analysis, nor did we replicate previously reported associations from individual cohorts. However, SNP-level summary statistics afford the best-available molecular genetic index of PTSD and can be used in polygenic risk prediction and genetic correlation studies . Furthermore, our unique data contributes to the broader goal of increased ancestral diversity in genomic data resources. In sum, results demonstrate marked heritability for the development of PTSD in women, shared genetic risk between PTSD and other psychiatric disorders, and highlight the importance of multi-ethnic/racial samples.
18. PGC Major Depressive Disorder (MDD) Work Group Update
Prof. Cathryn Lewis, King's College London
The PGC MDD working group aims to use genome-wide studies in major depressive disorder (MDD) to identify and characterise its genetic contribution. Our initial analysis (PGC1) of 9k cases and 9k controls (Ripke et al., Molecular Psychiatry, 2012) found no genome-wide findings. PGC MDD has since expanded to 29 studies of 16 823 cases and 25 632 controls. This is supplemented by summary statistics from external studies from iPSYCH, UK Biobank, GERA, deCODE, Generation Scotland, giving a total of 51 865 cases and 112 200 controls. Meta-analysis with summary statistics from the 23andMe MDD study (Hyde et al., Nature Genetics, 2016) is in progress.
Current analysis identifies seven loci at genome-wide significance, in six genomic regions, including NEGR1, LRFN5 and the HLA region. Expansion with published 23andMe results indicates that further regions will be identified in a full meta-anlaysis.
Genetic studies in depression have always been challenging, due to its high prevalence and modest heritability (37%). Our current studies confirm that a substantially increased sample size allows MDD to finally show its elusive genetic underpinnings.
19. CLOZUK: Update on GWAS for Schizophrenia
Dr. James Walters, University of Cardiff
Schizophrenia is a debilitating psychiatric condition often associated with poor quality of life and decreased life expectancy. Lack of progress in improving treatment outcomes has been attributed to limited knowledge of the underlying biology. We report the largest single cohort genome-wide association study of schizophrenia (11,260 cases and 24,542 controls) and through meta-analysis with existing data we identify 50 novel GWAS loci. We show for the first time that the common variant association signal is highly enriched among genes that are intolerant to loss of function mutations. We demonstrate that common schizophrenia risk variants persist in the population, despite the low fecundity associated with the disorder, through the process of background selection. Associations point to novel areas of biology (e.g. metabotropic GABA-B signalling and acetyl cholinesterase), reinforce those implicated in earlier GWAS studies (e.g. calcium channel function), converge with earlier rare variants studies (e.g. NRXN1, GABAergic signalling), identify novel overlaps with autism (e.g. RBFOX1, FOXP1, FOXG1), and support early controversial candidate gene hypotheses (e.g. ERBB4 implicating neuregulin signalling). We also demonstrate the involvement of six independent central nervous system functional gene sets in schizophrenia pathophysiology.
20. Pervasive Genetic Influence on Transcriptome Dysregulation Shared Across Major Psychiatric Disorders
Dr. Michael Gandal, UCLA
The predisposition to neuropsychiatric disease involves a complex, polygenic, and pleiotropic genetic architecture. However, little is known about how genetic variants impart brain dysfunction or pathology. We use transcriptomic profiling as an unbiased, quantitative readout of molecular brain-based phenotypes across 5 major psychiatric disorders, including autism (ASD), schizophrenia (SCZ), bipolar disorder (BD), depression (MDD), and alcoholism (AAD), compared with carefully matched controls. We identify clear patterns of shared and distinct gene-expression perturbations across these conditions, identifying neuronal gene co-expression modules downregulated across ASD, SCZ, and BD, astrocyte modules most prominently upregulated in ASD and SCZ, and microglia and signaling modules respectively observed in ASD and MDD. Remarkably, the degree of sharing of transcriptional dysregulation was strongly related to polygenic (SNP-based) overlap across disorders, indicating a significant genetic causal component. These findings provide a systems-level view of the neurobiological architecture of major neuropsychiatric illness and demonstrate pathways of molecular convergence and specificity.
21. Bromodomain Containing 1 (BRD1) in Psychiatric Disorders: from Genetic Association to Disease Biology Using Brain Transcriptomic Profiling in Genetically Modified (Brd1+/-) Mice
Dr. Per Qvist, PGC iPsych, Aarhus University
Post-translational modifications of histones are important in the pathophysiology of psychiatric disorders. Whereas histone acetyltransferases and deacetylases have been extensively explored as drug targets, less research has looked into readers of the histone code. Bromodomain-containing proteins serve as scaffolds in histone-, and chromatin modifying complexes and evidence supports their implication in neurodevelopmental disorders. Particularly, Bromodomain containing 1 (BRD1) has repeatedly shown genetic association with both schizophrenia and bipolar disorder and its interactome is significantly enriched with components implicated in neurodevelopment and mental illness. Underlining the importance of BRD1 in mental health, hampered Brd1 expression in genetically modified (Brd1+/-) mice, manifest as general and sex-specific behavioural-, brain morphometric-, and neurochemical deficits with broad translational relevance to psychiatric disorders. Supported by extensive transcriptomic profiling of multiple brain tissues in Brd1+/- mice, our data support a model in which BRD1 acts as a regulatory hub protein in mental health by facilitating dynamic transcriptional control in the developing-, and mature brain. Particularly, we provide novel evidence that links BRD1’s function to nuclear receptor mediated signalling, which may have implications in the treatment of psychiatric disorders and their gender-biased symptom profiles.
22. From GWAS to Function
Dr. Danielle Posthuma, PGC
Genome-wide association studies have yielded numerous genetic risk loci associated with mental disorders. However, risk loci detection is only the first step in understanding the biology of mental disorders. A statistically significant disease - gene association generates hypotheses that need to be tested in vitro for functional validation. That is, we need to show how genetic differences lead to a measurable phenotypic effect that is relevant for the disorder. We will provide examples of how to bridge the gap from GWAS to function, using in-silico techniques as well as wet-lab techniques from stem cell research.
23. Using iPSC-derived neurons for drug discovery in the CNS
Dr. Ricardo Dolmetsch, Novartis
24. Circuit-Based Investigation of CaV1.2 Hyperfunction in Symptom Domains Associated with Schizophrenia and Bipolar Disorder
Dr. Benjamin Hall, ROCHE
L-type calcium channels (LTCCs) are critical regulators of multiple cellular processes in neurons. Genetic variation in LTCC encoding genes, in particular CACNA1C, the gene coding for the channel forming alpha subunit of CaV1.2, have been repeatedly identified as conferring risk in schizophrenia (SZ) and bipolar disorder (BD). However, the causal links between CACNA1C genetic variation and symptom domains related to these disorders remain poorly understood. Evidence has accumulated showing increased expression of CaV1.2 in neurons derived from human cells and in post mortem analysis of brains from patients containing the risk-associated single nucleotide polymorphism (SNP) rs1006737. Additionally, heterozygous deletion of Cacna1c in mice has protective effects in measures of mood-related behaviors as well as in age-associated cognitive decline. In spite of this evidence, preclinical animal models that examine the consequence of CaV1.2 hyperfunction in individual neurons and specific circuits in vivo are lacking. To investigate this further, we have generated a humanized-in-brain, conditionally-expressed, inactivation mutant of the Cacna1c gene in mouse. This genetic strategy will allow us to determine consequences of mutant CaV1.2 expression in a circuit-specific manner in vivo using genetic and virally encoded Cre recombinase strategies.
25. Transcriptional Networks in Autism Spectrum Disorder
Dr. Daniel Geschwind, UCLA
Genome‐wide association and genome‐sequencing studies have led to identification of many loci for neuropsychiatric disorders, and in parallel, indicated that these conditions are remarkably genetically heterogeneous. Tissue specific transcriptomic data across multiple brain regions and key developmental epochs, as well as single‐cell‐type transcriptome data provide new opportunities for identifying and interpreting these complex genetic data in the context of robustly defined transcriptional networks. This talk will illustrate how construction of robust and reproducible transcriptional networks provides a powerful framework for beginning to understand the mechanisms underlying autism spectrum disorder.