Categories
BashTheBug on the Zooniverse News

The Aim

BashTheBug is an online project studying, with the help of the public, antibiotic resistance in Tuberculosis (TB). It will be launched  on the well-known and successful citizen science platform, Zooniverse.org, in April 2017.

As shown above, the website will show volunteers an image of a series of small, circular wells. Each well contains a sample of a strain of M.tuberculosis (the bacteria, or bug, that causes TB). The top two wells have no antibiotic and hence give you an idea of how well this strain grows. The bottom series of, in this case, six wells contain an increasing dose of antibiotic as you move from left to right. The volunteers are asked to say how well the TB grows in the presence of the antibiotic and whether there is a dose that kills (bashes?) the bug.

The BashTheBug project is part of a much larger global project, CRyPTIC, that is collecting 100,000 different TB strains from across Africa, Asia, Europe and the Americas at more than a dozen centres between now and 2020. Each strain will:

  1. Have its whole genome sequenced. This is comparatively cheap since the genome of TB is only 4 million bases, which is less than 0.1% the size of a human genome.
  2. Be tested against a panel of 15 different anti-TB drugs, ranging from front-line, established therapies, to more recent antibiotics.

The second part is being done using a specially-designed 96 well micro titre plate. Part of the TB sample is injected into the centre of each of the 96 wells and then the plate is incubated for 2-3 weeks to allow the bacteria to grow. As shown above, two of the wells have no antibiotic impregnated so the growth of the bacteria should be unimpeded. Below is an example of such a plate after 3 weeks of growth.

As you can see many of the wells are clear, indicating that the dose of antibiotic within them is sufficient to kill the bacteria. It would seem a comparatively easy task to determine whether bacteria are growing in each well, however, it is difficult for three reasons. Firstly the sheer number of measurements required is huge (if each plate is measured at two time points: 96 x 100,000 x 2 = 19.2 million well/measurements). Secondly, and this follows on from the first in a way, the volume of plates means we can only get a single scientist to examine each plate at each of the dozen or so global centres. Thirdly, there is an element of subjectivity to this task, especially when the strain does not grow very well (the photo above is a good clear example).

Hence, in addition to the expert evaluations, we are investigating automatic growth detection by computer and crowd-sourcing a consensus through the BashTheBug project. Whilst promising, the automatic growth detection algorithm is easily confused by artefacts in the image, like air bubbles, shadows and condensation. It is our hope and expectation that by combining the results of all three methods we will produce a much more accurate phenotypic dataset than would otherwise have been possible.

This is really, really important because the next part of the CRyPTIC project is putting the genetic and phenotypic (drug resistance) datasets side-by-side and using sophisticated statistical techniques to infer which mutations in which genes confer resistance to specific antibiotics (and equally importantly, which do not). Any significant errors in the phenotypic data could lead to spurious results which would take us time to run down. Of course, we know already many of the mutations that confer resistance, however the aim of the CRyPTIC project is to produce a comprehensive catalogue of mutations in the TB genome that confer resistance.

What is underlying all this research is a shift in clinical microbiology towards using genetic methods to diagnose bacterial disease and advise doctors which antibiotics are best able to treat a particular case. So, instead of a traditional culture-based method, where a sample from the infection is grown and then split, with each portion tested against a certain antibiotic to see which works and which doesn’t (much like the 96 well plate), the approach is to simply sequence the whole genome of the infecting pathogen and then, by looking up the mutations in a catalogue, infer which antibiotics would work and which would not. Because M.tuberculosis is so slow growing, the traditional approach can take 4-8 weeks to return a report to the doctors. Since this genetic-based approach already only takes 5-7 days, there is a huge clinical benefit for Tuberculosis.

This might sound like science-fiction, but it has already started happening in England. In March 2017, Public Health England rolled out routine whole genome sequencing for all new suspected cases of TB and it is likely other countries will follow. The more comprehensive catalogue that the CRyPTIC project will deliver, helped in no small part by the BashTheBug project, will further improve this approach.

We plan to launch BashTheBug as a Zooniverse.org medical project in April 2017. If you’d like to be informed when it launches, please add your email address to the form at the bottom of this page.

BashTheBug is part of the CRyPTIC project which is funded by the following organisations

 

By Philip Fowler

Philip W Fowler is a computational biophysicist studying antimicrobial resistance working at the John Radcliffe Hospital in Oxford.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.