Biology subjects

Melamed, R. D.

Publications and source records attributed to Melamed, R. D..

5 recordsLinked to original sources

bioRxiv2025

Discovering disease genetic variation impacting gene expression in 103 brain tissues with the Brain Ontology Expression (BRONTE) graph neural network model

Uncovering genes dysregulated in neuropsychiatric diseases promises to put forward therapeutic targets. By linking disease risk variants to tissue-specific gene regulation, the Genotype-Tissue Expression (GTEx) project has enabled discovery of disease genes and the brain tissues where these pathogenic effects occur. But, GTEx requires huge investment in obtaining human brain tissues, limiting the sample size and tissue diversity available. Crucially, this resource does not contain all key tissues where brain diseases are known to originate. Here, we propose a novel graph convolutional model that integrates the GTEx brain expression data with the Allen Brain Atlas, a resource spanning over 100 fine grained brain tissues that are annotated with an ontology of tissue organization. Our model, Brain Ontology Expression (BRONTE), allows completion of the GTEx data for 221 individuals to impute expanded expression in 103 uncollected detailed brain tissues. We show that expression data imputed by BRONTE recovers measured variation in expression across subjects, genes, and brain tissues. Most importantly, our method both recapitulates known relationships between genetic variations and gene expression, and simultaneously discovers both new genes impacted by genetic variants, and the tissues impacted. This allows us to put forward novel mechanisms of effect for genetic variation associated with 30 brain-related traits, supplying therapeutic targets in overlooked tissues of the brain.

bioinformatics↗

bioRxiv2023

Mendelian disease comorbidity as a resource for novel therapeutic targets

Drugs targeting disease causal genes are more likely to succeed for that disease. However, complex disease causal genes are not always clear. In contrast, Mendelian disease causal genes are well-known and druggable. Here, we seek an approach to exploit the well characterized biology of Mendelian diseases for complex disease drug discovery, by leveraging evidence of pathogenic processes shared between monogenic and complex disease. One way to find shared disease etiology is clinical association: some Mendelian diseases are known to predispose patients to specific complex diseases (comorbidity). Previous studies link this comorbidity to pleiotropic effects of the Mendelian disease causal genes on the complex disease. In previous work studying incidence of 90 Mendelian and 65 complex diseases, we found 2,908 pairs of clinically associated (comorbid) diseases. Using this clinical signal, we can match each complex disease to a set of Mendelian disease causal genes. We hypothesize that the drugs targeting these genes are potential candidate drugs for the complex disease. Our analysis shows that the candidate drugs are enriched among currently investigated or indicated drugs for the relevant complex diseases (odds ratio=1.84, p=5.98e-22). By combining comorbidity with genetic similarity, we recommend drugs further enriched for those investigated or indicated. Our findings suggest a novel way to take advantage of the rich knowledge about Mendelian disease biology to improve treatment of complex diseases.

bioinformatics↗

bioRxiv2023

Integrating disease genetics and drug bioassays to discover drug impacts on the human phenome

Unintended effects of medications on diverse diseases are widespread, resulting in not only harmful drug side effects, but also beneficial drug repurposing. This implies that drugs can unexpectedly influence disease networks. Then, discovering how biological effects of drugs relate to disease biology can both provide insight into the biological basis for latent drug effects, and can help predict new effects. Rich data now comprehensively profile both drug impacts on biological processes, and known drug associations with human phenotypes. At the same time, systematic phenome-wide genetic studies have linked each common phenotype with putative disease driver genes. Here, we develop Draphnet, a supervised linear model that integrates in vitro data on 429 drugs and gene associations of nearly 200 common phenotypes to learn a network connecting these molecular signals to explain drug effects on disease. The approach uses the -omics level similarity among drugs, and among phenotypes, to extrapolate impacts of drug on disease. Our predicted drug-phenotype relationships outperform a baseline predictive model. But more importantly, by projecting each drug to the space of its influence on disease driver genes, we propose the biological mechanism of unexpected effects of drugs on disease phenotypes. We show that drugs sharing downstream predicted biological effects share known biology (i.e., gene targets), supporting the potential of our method to provide insights into the biology of unexpected drug effects on disease. Using Draphnet to map a drugs known molecular effects to their downstream effect on the disease genome, we put forward disease genes impacted by drug targets, and we suggest new grouping of drugs based on shared effects on the disease genome. Our approach has multiple applications, including predicting drug uses and learning about drug biology, with potential implications for personalized medicine. Author summaryMedications can impact a number of cellular processes, resulting in both their intended treatment of a health condition, and also unintended harmful or beneficial effects on other diseases. We aim to understand and predict these drug effects by learning the network connecting the biological processes altered by drugs to the genes driving disease. Our model, called Draphnet, can predict drug side effects and indications, but beyond prediction we show that it is also able to learn a drugs expected effect on the disease genome. Using Draphnet to summarize the biological impact of each drug, we put forward the disease genes impacted by drugs or drug targets. For instance, both anti-inflammatories and some PPAR-agonists share downstream effect on the cholesterol ester transfer protein (CETP), a gene previously experimentally supported as an effector of fenofibrate. Our approach provides a biological basis for drug repurposing, potentially accelerating clinical advances.

bioinformatics↗

bioRxiv2019

Capturing medical practice with indication embeddings

ObjectiveThe electronic health record is a rising resource for quantifying medical practice, discovering adverse effects of drugs, and studying comparative effectiveness. One of the challenges of applying these methods to health care data is the high dimensionality of the health record. Methods to discover effects of drugs in health data must account for tens of thousands of potentially relevant confounders. Our goal in this work is to reduce the dimensionality of the health data with the aim of accelerating the application of retrospective cohort studies to this data. Materials and MethodsHere, we develop indication embeddings, a way to reduce the dimensionality of health data while capturing information relevant to treatment decisions. We evaluate these embeddings using external data on drug indications. Then, we use the embeddings as a substitute for medical history to match patients, and develop evaluation metrics for these matches. ResultsWe demonstrate that these embeddings recover therapeutic uses of drugs. We use embeddings as an informative representation of relationships between drugs, between health history events and drug prescriptions, and between patients at a particular time in their health history. We show that using embeddings to match cohorts improves the balance of the cohorts, even in terms of poorly measured risk factors like smoking. Discussion and ConclusionUnlike other embeddings inspired by word2vec, indication embeddings are specifically designed to capture the medical history leading to prescription of a new drug. For retrospective cohort studies, our low-dimensional representation helps in finding comparator drugs and constructing comparator cohorts.

bioinformatics↗

bioRxiv2017

Patchwork of contrasting medication cultures across the USA

Health care in the United States is markedly heterogeneous, with large disparities in treatment choices and health spending. Drug prescription is one major component of health care--reflecting the accuracy of diagnosis, the adherence to evidence-based guidelines, susceptibility to drug marketing, and regulatory factors. Using medical claims data covering nearly half of the USA population, we have developed a framework to compare prescription rates of 600 popular drugs in 2,334 counties. Our approach uncovers geographically separated sub-Americas, where patients receive treatment for different diseases, and where physicians choose different drugs for the same disease. The geographical variation suggests influences of racial composition, state-level health care laws, and wealth. Some regions consistently prefer more expensive drugs, even when they have not been proven more efficacious than cheaper alternatives. Our study underlines the benefit of aggregating massive information on medical practice into a summarized and actionable form. We hope that our methodology and conclusions will guide policy measures for aligning prescriptions with best-practice guidelines.

bioinformatics↗