New approaches to analysing biodata
Features
- Author: Joanna Carpenter
- Date: 10 Mar 2015
- Copyright: Image appears courtesy of Getty Images
A set of new tools and research challenges are providing new insights into human disease by promoting collaboration and identifying causal relationships
Stephen Friend, founder of Sage Bionetworks, likes to tell the story of how he realized it was time to approach medical research differently.
Thirty-odd years ago, as a paediatric oncologist, he met a new patient. The child’s father came too, and both father and son had their right eye missing as a result of a rare inherited eye cancer.
Despite progress in understanding genes since then, Friend says treatments have fallen behind what we know about genetic disorders.
Data-driven
Since 2009, Sage Bionetworks, the non-profit research organisation that Friend co-founded with his colleague Eric Schadt, has been working to bring novel ways to analyse biodata modeling for understanding and treatment of human disease.
It promotes data-driven predictive modelling by encouraging open science and collaboration, using data sharing, cloud computing and open social media.
Sage argues that the large quantities of data now available – together with these tools – are making progress on complex problems possible, without reducing them to a piecemeal set of smaller problems with lab-based solutions that fail in real life.
Sage Bionetworks has several ongoing projects that each approach this principle from a different angle.
Open platforms for data-sharing and collaboration
For instance, Sage has developed two open platforms for data: BRIDGE and Synapse.
BRIDGE is being built, with support from the Robert Wood Foundation, to allow patients to partner in the health research that matters to them, providing data from mobile sensors such as smartphones but also their insights on health problems.
Synapse is to help biomedical data scientists and clinical and biological researchers find and use data, to understand analysis workflows, to support genome-scale analysis, and to form collaborations. Synapse is being used by Sage as a shared working environment for its so-called DREAM (Dialogue for Reverse Engineering Assessments and Methods) Challenges to find causal relationships in complex biomedical systems.
DREAM Challenges
DREAM Challenges are a series of collaborative competitions based on open science. They are run by a community of researchers from a range of organizations – universities, non-profits, companies, and pharmaceutical / biotechnology companies.
In a DREAM Challenge, the organizers issue a research question and relevant data. Teams then compete to find the best solutions to the questions within the set time. The best solutions may be presented at conferences and written up for publication in journals.
Causal relationships
As an example, in 2014 one DREAM Challenge involved analysing experimental data on protein signaling networks from cancer cells. Molecules and their interactions form a causal network. If molecule A influences molecule B, then changes to A will lead to changes to B.
Around seventy international collaborations were given data on the abundance of different molecules in cells over time and under different experimental conditions. They then devised computational methods to infer causal networks from the data. Some teams made use of additional biological knowledge.
To test the teams’ models, the challenge organisers kept back some of their data. The models’ predictions were compared with the `test’ data to identify the best models. On average, the models of teams using additional biological knowledge performed better than those not using such knowledge.
Current and future DREAM challenges
A current DREAM Challenge concerns olfaction prediction, that is, mapping the chemical properties of odours to predict how a given molecule will smell. The dataset is based on descriptions by 49 human volunteers of the smell (strength, pleasantness, and so on) of 476 different odours.
Another Challenge, open for pre-registration now (March 2015) is on to predict survival of prostate cancer patients based on their clinical data.
Resilience Project
A Sage Bionetworks cancer project, known as Resilience and run jointly with the Icahn School of Medicine at Mount Sinai in New York City, has been looking for helpful mutations that stop genetic diseases.
Usually, disease-causing mutations are studied, but Stephen Friend, one of Sage’s founders, realized that in some families there may be those he calls `unexpected heroes’, who have the genetic mutation for a childhood disease but who have not developed it.
In a March 2014 TED talk, he explained that by contacting owners of anonymised DNA samples, such as the Children’s Hospital in Philadelphia and 23andMe, they had been able to identify dozens of candidates for unexpected heroes.
Since then, Sage has launched the beta phase of the project, asking adult volunteers to register for their DNA to be analysed for genetic mutations that typically cause rare diseases.
Maybe, after all these years, Friend will find mutation that protects its carrier from the rare eye cancer that first propelled him into research.
Connect: