A metabolomics1 breakthrough creates a fast, accurate and non-invasive way to screen for predisposition to an array of diseases and medical conditions.
Nuclear magnetic resonance (NMR) spectrometry can be used to determine the molecular composition of biofluids such as blood or urine, but identifying the type and concentration of specific metabolites is a time-consuming process with results dependent on consistent human interpretation. This limits the platform’s efficacy as an early-stage diagnostic tool for a wide spectrum of diseases.
Machine learning researcher Siamak Ravanbakhsh had never considered this problem when he started working toward a master’s degree at the University of Alberta. “Wouldn’t it be great if you could take some blood, push a button and see what’s in there?” his advisor, computer scientist Russell Greiner, asked one day. Ravanbakhsh got to work.
The ensuing breakthrough — a system called Bayesil that can quickly, accurately and automatically produce a metabolic profile from a small biofluid sample — is a big step toward the ability to predict the eventual onset of a number of diseases and medical conditions. Cost-effective and non-invasive, this technology could create myriad new applications for the science of metabolomics.
“We now have a toolbox to effectively analyze and interpret a complex set of compounds,” says Dr. Ravanbakhsh, who earned his master’s and PhD at the U of A and is currently a postdoctoral fellow at Carnegie Mellon University in Pittsburgh. “It’s not something for the future — it’s happening.”
Dr. David Wishart, one of Ravanbakhsh’s PhD advisors and leader of the Metabolomics Innovation Centre in Edmonton, says the research that led to this breakthrough started around 2000 as part of a broader effort to make metabolomics more mainstream. He uses an analogy to explain the significance of the advance: “We’ve been looking at the world through a keyhole, and now we’re looking through a picture window.”
By assessing the concentration of 50 to 60 “metabolic markers” such as glucose and glutamine, this system will be able help doctors ascertain predisposition to diseases such as diabetes and Alzheimer’s, as well as many cancers and even obesity. Early identification can prompt lifestyle or medical interventions that forestall the onset of health problems.
“Most of the exciting stuff in science happens when the fringes of fields like metabolomics and machine learning intersect,” says Dr. Wishart, noting that an Edmonton company has developed a urine test to screen for pre-cancerous colon polyps, and that a kit allowing labs to use the technique at the heart of Bayesil could be on sale by summer 2016. “In these new frontiers, collaborations can be very fruitful.”
1. The study of the set of metabolites present within an organism, cell or tissue.
2. Genomics and related disciplines such as proteomics, metabolomics and bioinformatics.
