Up to exabytes of data are potentially available from public sources but the biomedical sciences remain largely siloed into well-demarcated strata. In fact, the hierarchical organization of biological complexity can be represented as a multi-layered chart, in which each layer represents a domain of knowledge.

Unfortunately, little if any integration is realized across disciplines either experimentally, analytically or conceptually. We argue that integration of information from seemingly disparate disciplines such as genetics, pharmacology, physiology, and cellular biology can facilitate the inference of logical, and biologically plausible novel relationships. Indeed, if enough information is connected via biologically relevant paths into a giant graph, new knowledge will be produced as an emergent property of such an ensemble.

Biological Complexity
Data integration across disciplines in the biomedical sciences. SPOKE integrates data from more than 20 publicly available databases containing information ranging from molecular (sub-cellular) to organismal (multi-cellular) knowledge.

While integration and mining of large public datasets could prove a powerful approach to novel discoveries, these databases are often built using different architectures and standards, which hampers their straightforward integration

To tackle this problem, we created a scalable precision medicine open knowledge engine (SPOKE)—a giant heterogeneous graph currently boasting more than 3 million nodes and 5 million edges. SPOKE can be utilized to prioritize new uses for existing drugs (drug repurposing), to predict molecular targets of compounds and to create individualized patient profiles from electronic health records, among other applications. Current SPOKE projects at the Baranzini lab include:

Drug Repurposing for Progressive MS

Our lab is a partner in the BRAVEinMS consortium, a Progressive MS Alliance (PMSA)-sponsored collaboration with the goal of developing an effective therapeutic for progressive forms of MS using a drug repurposing approach.

Embedding of electronic health records (EHR) for precision medicine: We are developing approaches to integrate individual-specific medical information (from EHR) with population-specific knowledge (from SPOKE). The result of these approaches is a detailed “health barcode”, a map of the most biologically relevant variables within SPOKE for a particular patient, at a particular point in time. These barcodes can be utilized to group patients with similar characteristics, predict responders to therapeutic drugs, predict outcomes, understand mechanism of action of therapeutic drugs and more.

SPOKE: Continual Development

We are constantly adding more databases to SPOKE in order to enhance and its usefulness and broaden the scope of problems that could be tackled.

Additional SPOKE projects are underway at the Keiser and Bandyopadhyay labs at UCSF. 

Spoke