Social networks and genomes join forces

Tracking transmission: Scientists used social-network analysis to find the origins of an outbreak of tuberculosis (top). A patient designated MT0001 was thought to be ground zero for the outbreak, with other patients represented as circles. After sequencing bacteria genomes, scientists could track how the microbes moved from person to person (bottom), and discovered that there were two independent outbreaks. Credit: New England Journal of Medicine

I read an interesting article today on ‘Technology Review’ titled ‘Social Networking’s Newest Friend: Genomics’. It describes a recently published study on the emergence and spread of TB in British Columbia. In order to pinpoint the source of the disease, scholars did not only trace whole-genomes of the microbes responsible, but they combined this with a survey of the affected medium-sized community. By mapping possible interactions between individuals and examining DNA sequences attested, they were able to track the disease back to two independent sources.

This is a clear example of how social networks can be relevant “in real-time” as the data becomes available, to solve real problems. If the sources of such diseases can be identified early on, then officials and the community can take measures to prevent it from spreading even more. It is generally accepted in epidemology that human networks are media for the spread of disease and network approaches have been very popular to understand such processes. By combining a networks approach with genomics, however, an innovative and extremely detailed picture can be painted of a disease’s passage through a community.

This research is not an exception to commonly accepted issues surrounding social network analysis, however. Although I do not doubt the researchers did a thorough survey of the population focusing on a diversity of parameters to construct their networks, the limitation of types of relationships to those that we think might be influential as well as the formalisation/quantification of such relationships remain a necessary evil. It is very hard with such an approach to stumble upon unconnected clusters or parameters that were not thought to be of influence, for example. Basic sampling issues. Also, the construction of a thoroughly qualified social network takes time! I very much doubt that such an approach can be performed at the same speed as the spread of many modern-day diseases.

Having said that, this is a beautiful example of how two largely unrelated perspectives can lead to a completely new approach that enhances the results of both.

Advertisement