To build the first predictive simulation of MS, we have assembled and integrated existing demographic information, brain imaging data, long-term information on disease course and clinical response to treatments, and changes in gene activity before and after treatment from 3 databases comprising 9,000 people with MS.
This flagship program had the following major objectives:
Establish a novel alliance business model that expedites delivery of results and incentivizes partners and innovation.
Develop a data-sharing platform with integrated MS data for data-mining
Model 1.0: Build first-generation Computational & Simulation Disease Models
Progress Report: Orion MS Bionetwork
An innovative cooperative alliance model was established with the following Bionetwork Partners:
Accelerated Cure Project for MS, Brigham and Women’s Hospital, Exaptive, GNS Healthcare, MetaCell, Thomson Reuters.
Sponsorship was provided by Janssen Pharmaceuticals and additional funding provided by Philanthropic donations.
- Big data to smart data in Alzheimer’s disease: Real-world examples of advanced modeling and simulation; Magali Haas, Diane Stephenson, Klaus Romero, Mark Forrest Gordon, Neta Zach, Hugo Geerts, on behalf of the Brain Health Modeling Initiative; Alzheimer’s and Dementia 12 (2016) 1022-1030
- Big data to smart data in Alzheimer’s disease: The brain health modeling initiative to foster actionable knowledge; Hugo Geerts, Penny A. Dacks, Viswanath Devanarayan, Magali Haas, Zaven S. Khachaturian, Mark Forrest Gordon, Stuart Maudsley, Klaus Romero, Diane Stephenson, on behalf of the Brain Health Modeling Initiative (BHMI). Alzheimer’s and Dementia 12 (2016) 1014-1021
De-identified data from the following databases were curated and loaded into a cloud-based data knowledge management system called TranSMART™.
- Brigham and Women’s Hospital (BWH): The Comprehensive Longitudinal Investigation of MS at the Brigham and Women’s Hospital (CLIMB) dataset comprising over 2,000 patients with demographic, imaging, treatment and clinical measurements, including:
- Genome-wide association study (GWAS) data from ~1,200 patients
- Whole blood pre- and post-treatment gene expression changes from ~400 patients
- Longitudinal post-treatment gene expression changes from ~100 patients
Accelerated Cure Project (ACP): The ACP Repository contains over 3200 patient records from individuals with Multiple Sclerosis and other demyelinating disorders including neuromyelitis optica, transverse myelitis, acute disseminated encephalomyelitis, and optic neuritis.
- Over 50 clinical and covariate measures documenting clinical history and treatment were captured using a patient self-reported questionnaire;
- The database also includes almost 300 million molecular data points returned, to date, to the Repository database from the many academic and industrial research groups that have undertaken research with ACP’s Repository biosamples and data.
- PatientsLikeMe (PLM): Deep phenotyping of 4,000 patients derived from their 35,000-patient MS community that provides real-time, real-world outcome data beyond the clinic.
- Additionally, analysis of the phenomics of MS was conducted by PLM on their entire community database and compared to BWH CLIMB cohort.
Biomodeling & Simulation
Data-Driven MS Model: In partnership with GNS Healthcare, we used their powerful REFS Platform on the BWH dataset we built a SNP-only and SNP-RNA model of Multiple Sclerosis using Causal Inference/Machine Learning approaches.
Knowledge-Driven MS Model: In partnership with Thomson Reuters we utilized MetaBase, a deeply curated database of the existing literature to build molecular pathway models of Multiple Sclerosis.
Prognostic MS Phenomics Model: In partnership with PLM, we harnessed their database of >35,000 real-world patient-reported outcome data to build a prediction algorithm for MS disease course. The database was also analyzed to compare results from patient-reported outcomes to findings from published clinical data studies. These results will inform Orion MS 2.0 program activities.
C. Elegans Modeling: Orion is partnered with MetaCell to support OpenWorm, in order to seed the systems biology simulation approach it is taking in MS.
- What is OpenWorm?
OpenWorm is an open-source project that is building a whole-organism computer model of a C. elegans nematode. Construction of OpenWorm is being carried out by a non-incorporated community of volunteers.
OpenWorm has the ambitious goal of combining a physics simulation with a neuronal activity simulation to connect scientific knowledge about the worm’s cellular activity to the understanding of its physical behavior. The community is constantly expanding through publication of journal articles and recruitment of new members.
The algorithms and simulation approaches developed for the ‘simple worm’ will provide building blocks for modeling more complex systems such as the human.
Virtual Cell Project: MetaCell has also received support to create software specifications and an analysis document that will defines a development plan and architecture for a Virtual Cell to be developed through Orion Bionetwork’s 2.0 program.
- What is a virtual cell?
The VirtualCell project has the goal of creating a computer model of a prokaryotic cell using the totality of knowledge amassed on its components and biophysical processes. The model, which consists of multiple algorithms grouped in 4 major areas (DNA, RNA, metabolites and proteins), aims at predicting phenotype from genotype with unprecedented accuracy. In other words, dynamic simulations of a virtual cell would demonstrate all the behaviors of an actual cell.
- Scientists are convening across disciplines to address the challenges of engineering a virtual cell (i.e. “Towards the 3D virtual cell” at UCSD, December 2012). To date, a whole-cell computational model has been created for a small parasitic bacterium having the smallest known genome that can constitute a cell. This model was able to predict the phenotype from genotype with 80% accuracy (Karr et al)
- This first model is an important building block, from which other cell types can begin to be modeled. Eventually, it will be possible to create models of human cell types that are relevant to disease. A whole-cell model of a neuron will give us fundamental understanding of normal neuronal function, and, even more critically, the ability to probe what goes wrong in disease conditions like multiple sclerosis.
Data Mining: ACP is receiving support from Orion to work with the experts from the Alliance and undertake data-mining activities on its own repository in search of new knowledge that will lead to improved treatments and cures for MS and related disorders.
Simulation Platform: Orion is funding MetaCell to build the a Systems Biology Open Simulation Platform called Geppetto.
Geppetto is a multi-algorithm, multi-scale simulation platform engineered to support simulation of complex biological systems and their surrounding environment:
Geppetto allows separation of functionality into independent, interchangeable modules such that each contains everything necessary to execute a given aspect of desired functionality.
Geppetto can handle a growing amount of work in a robust fashion by being intrinsically distributed, scaling up to accomodate growing load.
Geppetto allows for future growth by including hooks and mechanisms for expanding/enhancing the system with anticipated capabilities, without having to make ad-hoc changes to the system infrastructure.
Geppetto is not tied to any specific biological simulation, nor to the model being simulated or the simulation aspects (neuronal, mechanical, etc.) being simulated.
Geppetto is based on a client-server model, where the simulation is controlled by a client through a web interface.
Geppetto architecture needs to allow separation of the execution of a simulation into multiple processes which can be executed by different server and which communicate with each other by exchanging messages.
- Dynamic Driven
Geppetto components can be deployed, re-deployed, and un-deployed without a system (server) restart.
Visualization Platform: In partnership with Exaptive, we are supporting a proof-of-concept project that will explore the CLIMB/ACP datasets in tranSMART, push forward tranSMART integration and visualization through the generation of open-source libraries, using the Exaptive platform.