AI Algorithm can Help Predict Bacterial Infections, Develop New Treatments

A better understanding of how bacterial proteins function together as a network to take control of human cells could help estimate the outcomes of infections and design novel treatments.

Image Credit: Pashkova/

Disease-causing bacteria, like Salmonella and E. coli, are similar to how a hacker takes control of an organization’s software to create chaos. These microorganisms use tiny molecular syringes to administer their own chaos-inducing agents (known as effectors) inside the cells that keep the human gut healthy.

Such effectors seize control of the human cells, which overwhelm their defense systems and inhibit the crucial immune responses, thus enabling the infection to take hold.

In the past, studies have explored only single effectors. Now, a research team—headed by scientists from Imperial College London and The Institute of Cancer Research, London, which also included investigators from the United Kingdom, Israel, and Spain—has looked at the entire sets of effectors in various combinations.

The research work, recently published in the Science journal, analyzed data from experiments performed in mice. These animals were infected with the mouse model of E. coli, known as Citrobacter rodentium, which injects a total of 31 effectors.

The outcomes demonstrated how effectors function collectively as a network, allowing them to colonize their hosts even if certain effectors are eliminated. The study also showed how the immune system of the host can bypass the barriers created by the effectors, stimulating complementary immune responses.

Inherent Strength and Flexibility

According to the team, a better insight into the composition of the effector networks and their influence on the infections to take hold can help design new interventional strategies that interfere with their effects.

The data represent a breakthrough in our understanding of the mechanisms of bacterial infections and host responses. Our results show that the injected effectors are not working individually, but instead as a pack.

Gad Frankel, Study Lead and Professor, Department of Life Sciences, Imperial College London

We found there is an inherent strength and flexibility to the network, which ensures that if one or several components don’t work, the infection can go on. Importantly, this work has also revealed that our cells have a built-in firewall, which means that we can deal with the hacker’s corruptive networks and mount effective immune responses that can clear the infection,” Professor Frankel added.

Our study shows that we can predict how a cell will respond when attacked by different combinations of bacterial effector proteins .The research will help us to better understand how cells, the immune system and bacteria interact, and we can apply this knowledge to diseases like cancer and inflammatory bowel disease where bacteria in the gut play an important role.

Jyoti Choudhary, Study Co-Lead Author and Professor, Functional Proteomics Lab, The Institute of Cancer Research, London

Professor Choudhary added, “We hope, through further study, to build on this knowledge and work out exactly how these effector proteins function, and how they work together to disrupt host cells. In future, this enhanced understanding could lead to the development of new treatments.”

Predicting the Outcome of Infection

While performing the experiments, the researchers successfully removed different effectors when infecting the mice with the bacterium and monitored the effectiveness of each infection. This demonstrated that the network of effectors created by the microbe could be decreased by as much as 60%, and can still generate a successful infection.

The researchers obtained data on over 100 different synthetic combinations of the 31 effectors, which were used by Professor Alfonso Rodríguez-Patón and Elena Núñez-Berrueco from the Universidad Politécnica de Madrid to create an artificial intelligence (AI) algorithm.

The AI model successfully predicted the results of infection with Citrobacter rodentium that expressed different networks of effectors, and these effectors were tested with experiments in mice.

Since it was not possible to test all the potential networks formed by 31 effectors in laboratory settings, using an AI model provide the only practical method for investigating biological systems of this complexity.

The AI allows us to focus on creating the most relevant combinations of effectors and learn from them how bacteria are counteracted by our immune system. These combinations would not be obvious from our experimental results alone, opening up the possibility of using AI to predict infection outcomes.

Dr David Ruano-Gallego, Study Co-First Author, Department of Life Sciences, Imperial College London

Dr Julia Sánchez-Garrido, the co-first author of the study from the Department of Life Sciences at Imperial College London, stated, “Our results also mean that in the future, using AI and synthetic biology, we should be able to work out which cell functions are essential during infection, enabling us to find ways to fight the infection not by killing the pathogen with antibiotics, but instead by changing and improving our natural defence responses to infection.”

The study was funded by the Wellcome Trust.

AI analysis of how bacteria attack could help predict infection outcomes

Video Credit: Imperial College London

Journal Reference:

Ruano-Gallego, D., et al. (2021) Type III secretion system effectors form robust and flexible intracellular virulence networks Science.


Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type