Targeted drug therapies for bacterial infections

Researchers at the Helsinki Institute for Information Technology HIIT have made a computational breakthrough that will help the development of drug therapies targeted against bacteria causing serious infections. Genetics has a lot to gain from computational inference and statistical modelling!

In the last ten years, bacteria resistant to nearly all antibiotics have appeared all over the world. Bacteria are transmitted from continent to continent through global travel, and their resistance to antibiotics has been classified as a global threat to human health by the World Health Organization.

“A new method based on our research makes it possible to target drugs on certain bacterial genes in a way that will minimise the development of drug resistance and the possibility of interspecies transmission,” says Professor Jukka Corander of HIIT and the University of Helsinki.

Statistical modelling reveals evolutionary limitations in bacterial genetic variation

The research breakthrough, which has just been published in the respected and peer-reviewed journal PLOS Genetics, demonstrates the possibilities that this computational method offers to drug development.

The researchers analysed a large amount of pneumococcal and streptococcal bacteria samples and developed a statistical model that can be used to reveal evolutionary limitations in bacterial genetic variation occurring in the natural environment.

The research revealed several previously unknown epistatic mutation interactions in pneumococcal bacteria.

 “By combining data made available by these findings with molecular medicine, the development of new drug therapies can be started in order to disrupt the functioning of relevant bacterial genes and at the same time to block bacterial cell renewal,” says Jukka Corander.

”In most cases, epistatic interactions are closely tied to the core genome of a certain species of bacteria. Therefore, newly developed drugs would have no side effects on other bacteria.” 

According to Corander, this is the first time a probability model illustrating co-evolutionary pressure between mutations has been successfully adapted to all mutations present in a genome at the same time.

The model consists of nearly one hundred billion unknown variables

By utilising years of experience in algorithmic inference, the researchers developed an approach enabling a sufficiently accurate analysis.

According to Jukka Corander, in the near future the research results, based on comprehensive genome data, will provide an opportunity to study genetic variation in all common bacteria that cause infectious diseases. He believes that this offers basic research the chance to lay the foundation for the development of applications in precision medicine and promote the prevention of serious infections.

Jukka Corander is the director of the Computational Inference (COIN) research programme at HIIT and principal investigator of the recently published project. In his words, the computational challenge is enormous due to the fact that statistical models of this type contain nearly one hundred billion unknown variables.

The solution required several hundred thousand hours of CPU time provided by the CSC IT Center for Science server. Other important partners included researchers specialised in bacterial genomics at the Sanger Institute, a world leader in genome research.



Interacting networks of resistance, virulence and core machinery genes identified by genome-wide epistasis analysis. PLOS Genetics, DOI: 10.1371/journal.pgen.1006508.

PLOS Genetics:

Pictured: Jukka Corander.


Further information:

Professor Jukka Corander, University of Helsinki, Department of Mathematics and Statistics, Helsinki Institute for Information Technology HIIT, tel. +358 50 415 5294, email:

Minna Meriläinen-Tenhu, @MinnaMeriTenhu, 050 415 0316,