Dr Oliver Deusch – Gut feeling meets big data analysis

Certain bacteria could make you fat. That was the surprise finding by an American research team based at the Centre for Genome Studies in 2006 *1 who looked at the differences in gut bacteria between fat and thin people.  Specific groups of microbes were not just associated with obesity, they actually seemed to catalyse and thrive on it *2.

Discoveries about the intimate relationship between our body’s bacterial stowaways and our own health and wellbeing are becoming increasingly commonplace. In fact people are bacterial as much as human. This may sound surprising, but there are numerically more bacteria in the body than our own cells. The majority of these are in our intestines and we’ve evolved with these microbiomes (communities of bacteria) to such an extent that they are now massively important to normal bodily function.

If bacteria are associated with obesity in humans, it seems logical they may play a similar role in animals. That may not matter much in nature, but with obesity rising in the pet population, could commercial pet foods be tailored to positively affect the bacteria in their guts? This is a question Dr Oliver Deusch is keen to answer. A biologist by training he became intrigued by the emerging field of bioinformatics, the hybridization of computer science and life sciences.
“Originally we were limited to studying single bacterial genes” he says, “But advancements in DNA sequencing technology and computing mean we can now study all the genes in a bacterial cell.  This creates extraordinary potential to deliver new health solutions”.

We can now study all the genes in a bacterial cell. This creates extraordinary potential to deliver new health solutions”

His first application of this approach explored how high or low protein diets affect the bacteria in the guts of kittens as they grow *3. “We had already shown that we could accurately identify a huge range of bacteria. However, many were novel so we didn’t know what they were doing.”

To solve the problem Dr Deusch took a holistic metagenomic approach. That means looking at all the DNA in a sample, rather than just the regions used to identify individual bacteria.  By translating the DNA sequences into the protein functions they encode, the capability of a gut microbiome as a whole can be understood. “The beauty of this approach is you can see how the bacterial community functions as a whole, not just individuals” he says.


*1 Turnbaugh et al 2006. An obesity-associated gut microbiome with increased capacity for energy harvest
*2 Turnbaugh et al 2009. A core gut microbiome in obese and lean twins
*3 Deusch et al 2014. Deep Illumina-Based Shotgun Sequencing Reveals Dietary Effects on the Structure and Function of the Fecal Microbiome of Growing Kittens

Dr Deusch at WINSS 2013

That was the theory, the reality was much more challenging. Taking samples directly from kitten intestines would have been too invasive. Instead they worked with faecal samples since faeces collects intestinal bacteria as it transits through the gut.

The next obstacle was identifying the best approach to interrogate data on the terabyte scale. “We had to test multiple programs to find the most efficient algorithms, we finally arrived at one that was 300 times faster than the standard algorithm. Even then you still need massive computing power to interrogate the data” he says. Hence he turned to cloud computing, outsourcing the computing power required rather than invest in computers that would quickly outdate. “Ultimately the project used over 50,000 hours in processing time to assign bacterial functions to the data. The equivalent of running your home computer flat out for over a year“.

From the kittens on both high and low protein diets they identified over 7000 bacterial proteins. That was when the maths got complicated. Using a statistical approach they were able to shortlist over 2000 bacterial proteins that were significantly different when fed different diets. These numbers were still too large to make sense of what was happening.
“Bacteria can use the same protein function in different pathways so understanding which bacterial functions were affected by the diet was tricky”. The breakthrough came when they ran simulations on all the biological pathways and calculated which had significantly more protein functions affected by the high protein diet.

430px source shutterstock
“With such knowledge tailor made diets could be produced to select for these bacteria and improve pet health.”

Strikingly of the 10 most affected bacterial pathways, 6 were involved in amino acid metabolism proving his approach had worked. The main difference in diet was protein content and proteins consist of amino acids. What he had proved was that enough protein in the high protein diet had travelled to the gut and selected for bacteria that were most efficient at breaking down amino acids.


Having a metagenomics capability opens up many research directions; the effects of diet, age, obesity on the gut bacteria of pets can now be investigated.  “The next step is to understand what gut bacteria are associated with healthy cats and dogs versus, for example, obesity,” says Dr Deusch. “With such knowledge tailor made diets could be produced to select for these bacteria and improve pet health.”