How we are working towards prevention and diagnostics of Alzheimer’s, and drug treatments for AD, Parkinson’s and other diseases
Genes contribute to disease – Some genes more than others
We recognize that many genes are often involved in contributing to one disease. These genes work together in processes, and processes work together in unusual ways to cause disease.
We perform in silico functional modelling to interpret the activities of genes as sets of co-expressed functional pathways. This systematic functional framework allows for prediction of the key pathways that are correlated in their activity between a disease, a drug, and response to its application. Our approaches provide us with the ability to determine the dynamics of pathway activity and how they relate to each other. It’s a fundamental window into the function of the cell. It allows us to see alterations in pathway activity with disease progression – allowing us to map and predict which pathways will be affected and which pathways are driving sets of processes. Given these insights, this system is an obvious tool for helping us decide on which pathways and genes are important to consider for therapeutic intervention.
We take network models of gene-gene interaction, overlaid and organised according to interacting functional modules. Combined with appropriate ontology development, the interplay between metadata from experiments, functional characterisation, data integration and genome variant mapping and associated gene priorization provides a sound basis for integration. By making this resource available to the community, we work with a cadre of community researchers interested in building models from data.
Diseases and programmes we are working on
Together with Rudi Tanzi at Massachusetts General Hospital and the Centre for Health Bioinformatics group at Harvard E. Chan School of Public Health we are working in the Cure Alzheimer’s Genome Project to interpret the whole genome sequences of 1500 AD afflicted individuals, some in families. This has meant rethinking how to call variants in a realistic timeframe, while developing in an appropriate best practice framework. With that problem well along to being solved, we are now turning to linking called variants that have reached some level of significance within affected individuals (in families or across cohorts of affecteds) to the actual causative processes for the disease. We’re excited to link these data and to work with our collaborators in findings from the Accelerating Medicines Project – Alzheimer’s Disease – enabling functional association with genomic variants.
Resilience against Alzheimer’s
Because there are susceptible aged individuals who do not show any signs of Alzheimer’s, yet may have pathological hallmarks of Alzheimer’s – we are particularly interested in finding the basis of resilience against onset of Alzheimer’s. Funded by the National Institutes of Health, our project The Alzheimer’s Disease Resiliome: Pathway Analysis and Drug Discovery seeks to determine the key pathways that are differentially activated in resilient subjects, and, using drugs and miRNAs, emulate the activation of these pathways in a 3D organoid model of Alzheimer’s.
Human Experimental Models
A major problem in translating human genomic variation into targetable therapeutics is that to test targets, we have to use animal models. Animal cures are not human cures and this xenobiotic loss in translation can be a show stopper. So instead we turn the process on its head. We make molecular signatures from post mortem patient tissues. Then we look to see if genes interacting in and around these signatures turn up in human genetic studies. If they do, we consider the signatures associated with the disease at some level.
This means not only thinking about genes that are involved, but how they are interacting to contribute to onset and aetiology of AD. It’s a completely different kind of approach to the problem which explores the complex relationships amongst genomic variants and disease pathology. Our current approaches integrate the heterogenous sources of high throughput assays while attempting a synthesis that reflects the functional interactions of genes in pathways.
Together with a group of really motivated experts, (Lars Bertram, Joseph Ecker, Brad Hyman, Manolis Kellis, Rudolf Jaenisch, Andreas Pfenning, Rudy Tanzi and Li-Huei Tsai) we have been brought together by the Cure Alzheimer’s fund to leverage computational and laboratory tools that combine genetics, epigenomics, functional modeling and cellular phenotyping. We are building a map of the circuitry of Alzheimer’s-related genetic variants and relating the map to novel drug targets for Alzheimer’s. Our collaboration aims to result in some of the most comprehensive epigenomic and functional dissection of disease genes, variants, cell types and regulators. We integrate findings from this study into a platform providing infrastructure: the Cure Alzheimer’s Commons, an open access resource being co-developed together with Harvard researchers at the HSCI Centre for Stem Cell Bioinformatics to exploit growing resources of human sequence and functional genomics data in order to develop reproducible models for drug discovery. It provides a framework for functional and drug discovery translation of exomes and genomes of study participants with neurodegenerative diseases.
Mapping Genes to Pathways to Drugs and noncoding RNA
We use systems biology – networks of correlated pathway activities – to connect disease-associated gene activity in pathways to the response of pathways to drug interventions. We use this map to predict those drugs we think would be useful. We published this approach recently going after and successfully addressing sepsis in a mouse model. Now we are building noncoding RNA into our systems as drugs – highly specific – highly efficient with low side effects.
We have worked closely with industrial partners including Biogen Inc. (disclosure, I used to consult for Biogen) to help analyse and build models of neurodegenerative disease.
Translation of model systems to human health
Using an in silico approach that compares functional activity between model systems such as zebrafish, mouse and humans, we isolate key genes and pathways in the models that may be contributing to diseases in human subjects. We want to determine the dynamics of pathways as they alter their activity within disease progression, and to predict which subjects will respond to a drug treatment – and predict the drugs they will respond to.