GWGareth Williams

Gareth Williams

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Gareth Williams has a total of 211 co-authors. Here are the top co-authors by number of shared publications:

  • Patrick Doherty

    Wolfson Centre for Age-Related Diseases, King's College London, London, United Kingdom

    32 shared publications
  • Frank S. Walsh

    Ossianix, Ltd, Stevenage Bioscience Catalyst, Gunnels Wood Rd, Stevenage, Hertfordshire SG12FX, UK; Ossianix, Inc, University City Science Center, 3711 Market St, Philadelphia, PA 19014, USA

    10 shared publications
  • Clive Ballard

    Department of Clinical and Biomedical Science, Faculty of Health and Life Sciences, University of Exeter, Exeter, UK

    7 shared publications
  • Emma-Jane Williams

    Wolfson Centre for Age-Related Diseases King's College London SE1 1UL London United Kingdom

    7 shared publications
  • Patrick Doherty

    Wolfson Sensory Pain and Regeneration Centre, Institute of Psychiatry, Psychology and Neuroscience, King's College London, London, SE1 UL, UK

    7 shared publications
  • Clive Ballard

    University of Exeter Medical School, College of Medicine and Health, University of Exeter, UK

    6 shared publications

Latest Publications

  • Open Access

    Disease-Specific Perturbational Convergence Between Genetic-Risk and Disease-State Transcriptomic Signatures in Alzheimer's Disease

    by Gareth Williams

    Background Alzheimer's disease (AD) genetic susceptibility and disease-associated transcriptional changes are typically studied separately. Whether genetically inferred molecular mechanisms converge with transcriptional changes observed in post-mortem brain tissue remains unclear. Methods Cell-type-specific AD differential expression signatures were generated from five independent single-cell transcriptomic studies comprising over two million brain cells. In parallel, polygenic risk score (PGS)-associated expression signatures were inferred across 3,202 individuals using GTEx eQTL effect estimates. Both representations were compared at the individual gene, pathway, and drug-perturbational levels using large-scale perturbational profiles from Connectivity Map (CMap) and LINCS resources. Results AD single-cell signatures exhibited substantial reproducibility across cell types and correlated positively with bulk-brain pathology signatures. Genetically inferred signatures showed minimal concordance with disease-derived profiles at the level of individual genes and biological pathways. In contrast, there was a positive association when projecting both signatures onto drug-response space. Vasodilators, antihypertensives, and metabolic agents were enriched among compounds showing consistent reversal signatures across both disease representation. Conclusions In the present analysis, genetic risk and end-stage AD pathology are largely decoupled at the level of individual genes and pathways but converge on a shared, pharmacologically modifiable response landscape. This convergence is consistent with biological mechanisms linked to established vascular and metabolic dementia risk factors, suggesting perturbational profiling as a potential framework for future therapeutic investigation rather than a basis for immediate drug prioritisation.

  • Open Access

    Retrospective case-control study of pre-diagnosis observational and prescription data in Parkinson’s disease

    by Patrick Doherty, Susan Duty, Gareth Williams

    There are currently no disease-modifying treatments for Parkinson's disease. Retrospective analysis of routinely collected health data may enable earlier risk detection and identify therapeutic candidates for drug repurposing. We conducted a retrospective case-control study using data from the Clinical Practice Research Datalink. Individuals diagnosed with Parkinson's disease (n = 37,910) were matched by age, sex, and general practice to 113,730 controls. Biometric measures and prescription histories were analysed across two pre-diagnostic windows (5-10 years and 10-20 years before diagnosis). Logistic regression models were used to evaluate associations between biomarkers, medication exposure, and Parkinson's disease incidence. A composite risk score was constructed and assessed using receiver operating characteristic analysis. Individuals who developed Parkinson's disease showed significant deviations in established vascular, metabolic, and inflammatory markers years before diagnosis, including elevated systolic blood pressure, serum urea, and creatinine. The combined risk model demonstrated moderate predictive accuracy (area under the curve = 0.73). Previously reported contrasting associations of salbutamol and propranolol with Parkinson's disease incidence were confirmed. Hormonal therapies and triptan medications were associated with lower disease incidence but were more frequently prescribed to individuals with more favourable biometric risk profiles, suggesting potential confounding. In contrast, glucagon-like peptide-1 receptor agonists were prescribed to individuals with higher underlying risk yet were associated with reduced disease incidence. Integration of biometric and prescription histories improves interpretation of medication-disease associations in population data. These findings support glucagon-like peptide-1 receptor agonists as candidates for further investigation in Parkinson's disease based on observational associations and demonstrate the value of real-world data for early identification of population-level risk patterns and therapeutic discovery in neurodegenerative disease.

  • Open Access

    Identifying synergistic combinations of repurposed treatments for Alzheimer’s Disease

    by Clive Ballard, Janet Sultana, Pat Doherty, Gareth Williams, Anne Corbett


  • Open Access

    Retrospective Case-Control Study of Pre-Diagnosis Observational and Prescription Data in Parkinson’s Disease

    by Patrick Doherty, Susan Duty, Gareth Williams

    Background There are still no disease-modifying treatments available for Parkinson’s disease (PD) despite the growing interest in trialling repurposed drugs. Retrospective analysis of real-world health records may identify early predictive biomarkers and therapeutic candidates for increasing the chances of a positive trial outcome. Methods We conducted a retrospective case-control study using the UK-wide Clinical Practice Research Datalink (CPRD), comparing biometric and prescription data from over 38,000 individuals diagnosed with PD to 114,000 matched controls, across 5 to 10 and 10 to 20 years prior to diagnosis. Results The PD cohort was characterised by significant deviations in established biometric risk factors, including markedly elevated systolic arterial pressure, serum urea, and creatinine levels, alongside widespread alterations in metabolic and inflammatory markers. A combined risk score provided a moderate predictive accuracy (ROC AUC = 0.73) in line with published analyses. We confirmed the previously reported contrasting associations of salbutamol and propranolol with PD incidence. We further found hormonal drugs and triptan migraine medication associated with a lower PD incidence. However, these drugs had a higher prescription rate in those with lower biometric-based risk possibly confounding their positive effects against PD risk. In contrast GLP-1 receptor agonists tended to be prescribed in those with higher underlying PD risk but were associated with a lower PD incidence. Conclusion Integrating biometric and prescription histories enables more accurate interpretation of medication-disease associations, highlighting GLP-1 receptor agonists as strong candidates for PD intervention. Our findings support a data-driven framework for early risk detection and therapeutic discovery in neurodegenerative diseases.

  • Open Access

    DAGLα/β, 2-AG release, and Parkinson’s Disease: Exploring a causal link

    by Patrick Doherty, Gareth Williams

    The diacylglycerol lipases, DAGLα and DAGLβ, hydrolyse diacylglycerol (DAG) to produce 2-arachidonoylglycerol (2-AG), a key endocannabinoid (eCB) and CB 1 /CB 2 receptor ligand. While DAGLα is well established as a regulator of CB 1 -dependent synaptic plasticity, recent studies have identified DAGLB mutations as a cause of autosomal recessive early-onset Parkinson’s disease (PD). Here, we present a comprehensive analysis of DAGLβ mRNA expression, demonstrating its co-expression with DAGLα mRNA predominantly in excitatory neurons throughout the adult nervous system. We see no evidence for enrichment of the DAGLs or CB 1 transcripts in the striatum or in dopaminergic neurons. We discuss these findings within a review of recent literature that points to a wider involvement of the eCB system in PD. Notably, DAGLα-dependent 2-AG release at synapses relies on α-synuclein function—a protein central to PD pathophysiology—implicating both DAGLs in PD and pointing to widespread disruption in 2-AG release. Consistent with this, substantial reductions in 2-AG levels have been reported in the cerebrospinal fluid (CSF) of PD patients. Depression, a major non-motor symptom of PD, often precedes the onset of motor deficits by several years. Human and mouse genetic studies suggest that reduced DAGL activity may contribute to depression by impairing 2-AG-mediated CB 1 receptor signalling, which is crucial for synaptic plasticity, stress resilience, and mood regulation. These findings point to a potential causal link between DAGL dysfunction and the non-motor symptoms in PD.


  • Open Access

    Transcription-Driven Repurposing of Cardiotonic Steroids for Lithium Treatment of Severe Depression

    by Richard Killick, Claudie Hooper, Cathy Fernandes, Christina Elliott, Dag Aarsland, Svein R. Kjosavik, Ragnhild Østerhus, Gareth Williams

    Lithium is prescribed as a mood stabilizer for bipolar disorder and severe depression. However, the mechanism of action of lithium is unknown and there are major side effects associated with prolonged medication. This motivates a search for safer alternative drug repurposing candidates. Given that the drug mechanism may be encoded in transcriptional changes, we generated the gene expression profile for acute lithium treatment of cortical neuronal cultures. We found that the lithium-associated transcription response harbors a significant component that is the reverse of that seen in human brain samples from patients with major depression, bipolar disorder, and a mouse model of depression. Interrogating publicly available drug-driven expression data, we found that cardiotonic steroids drive gene expression in a correlated manner to our acute lithium profile. An analysis of the psychiatric medication cohort of the Norwegian Prescription Database showed that cardiotonic prescription is associated with a lower incidence of lithium prescription. Our transcriptional and epidemiological observations point towards cardiotonic steroids as possible repurposing candidates for lithium. These observations motivate a controlled trial to establish a causal connection and genuine therapeutic benefit in the context of depression.


  • Open Access

    Transcription driven repurposing of cardiotonic steroids for Lithium treatment of severe depression

    by Richard Killick, Claudie Hooper, Cathy Fernandes, Christina Elliott, Dag Aarsland, Svein R. Kjosavik, Ragnhild Djønne Østerhus, Gareth Williams

    Lithium is prescribed as a mood stabiliser in bipolar disorder and severe depression. However, the mechanism of action of Lithium is unknown and there are major side effects associated with prolonged medication. This motivates a search for safer alternative drug repurposing candidates. Given that the drug mechanism may be encoded in transcriptional changes we have generated the gene expression profile for acute Lithium treatment of cortical neuronal cultures. We found that the Lithium associated transcription response harbours a significant component that is the reverse of that seen in human brain samples from patients with Major Depression, Bipolar Disorder and a mouse model of depression. Interrogating publicly available drug driven expression data we found that cardiotonic steroids drive gene expression in a correlated manner to our acute Lithium profile. An analysis of the psychiatric medication cohort of the Norwegian Prescription Database showed that cardiotonic prescription is associated with a lower incidence of lithium prescription. Our transcriptional and epidemiological observations point towards cardiotonic steroids as possible repurposing candidates for Lithium. These observations motivate a controlled trial to establish a causal connection and genuine therapeutic benefit in the context of depression.

  • Open Access

    A Predictive Model for Compound-Protein Interactions Based on Concatenated Vectorization

    by Gareth Williams, Kaz Azim

    Background: Large data sets of compound activity lend themselves to building predictive models based on compound and target structure. The simplest representation of structure is via vectorisation. Compound fingerprint vectorisation has been successfully employed in predicting compound activity classes. Results: A vector representation of a protein-compound pair based on a concatenation of the compound fingerprint and the protein triplet vector has been used to train random forest and neural network models on multiple datasets of protein-compound interaction together with compound associated transcription and activity profiles. Results for compound-target predictability are comparable with more complex published methodologies. Conclusion: A simple intuitive representation of a protein-compound pair can be employed in a variety of machine learning models to gain a predictive handle on the activity of compounds for which there is no activity data. It is hoped that this transparent approach will prove sufficiently portable and simple to implement that drug discovery will be opened up to the wider research community.


  • Open Access

    Pre-Diagnosis Observational and Prescription History Associated with Alzheimer’s Disease Incidence

    by Gareth Williams

    Background Alzheimer’s disease (AD) has so far proved refractory to intervention. However, disease incidence is variable across prior medication and observational measures. The present study adopted a data-driven approach to inform possible drug repurposing strategies in the light of concurrent prescription and biometric data. Methods and Results A real-world dataset was harnessed to compare observational and prescription data for 250,000 individuals prior to AD diagnosis against an age-, sex-, and clinical practice-matched control cohort with no AD history. Observation data was shown to both explain the associations of classes of drug prescription with AD incidence and to bolster the repurposing potential of the GLP-1 agonist class of anti-diabetic drugs. Conclusion This study provides insights into how drug repurposing based on prescription histories can be informed by concurrent observational data. These findings offer novel insights to be explored in future research on causal models for AD progression.


  • Open Access

    Neurodegenerative Disease Associated Pathways in the Brains of Triple Transgenic Alzheimer’s Model Mice Are Reversed Following Two Weeks of Peripheral Administration of Fasudil

    by Richard Killick, Christina Elliott, Elena Ribe, Martin Broadstock, Clive Ballard, Dag Aarsland, Gareth Williams

    The pan Rho-associated coiled-coil-containing protein kinase (ROCK) inhibitor fasudil acts as a vasodilator and has been used as a medication for post-cerebral stroke for the past 29 years in Japan and China. More recently, based on the involvement of ROCK inhibition in synaptic function, neuronal survival, and processes associated with neuroinflammation, it has been suggested that the drug may be repurposed for neurodegenerative diseases. Indeed, fasudil has demonstrated preclinical efficacy in many neurodegenerative disease models. To facilitate an understanding of the wider biological processes at play due to ROCK inhibition in the context of neurodegeneration, we performed a global gene expression analysis on the brains of Alzheimer’s disease model mice treated with fasudil via peripheral IP injection. We then performed a comparative analysis of the fasudil-driven transcriptional profile with profiles generated from a meta-analysis of multiple neurodegenerative diseases. Our results show that fasudil tends to drive gene expression in a reverse sense to that seen in brains with post-mortem neurodegenerative disease. The results are most striking in terms of pathway enrichment analysis, where pathways perturbed in Alzheimer’s and Parkinson’s diseases are overwhelmingly driven in the opposite direction by fasudil treatment. Thus, our results bolster the repurposing potential of fasudil by demonstrating an anti-neurodegenerative phenotype in a disease context and highlight the potential of in vivo transcriptional profiling of drug activity.