A few people who come to this blog know that I’ve been commercially active with thingslearn Ltd. in the UK. The company served as a port-of-call for customers mostly in the IoT and M2M arena, who were interested in the expert data science and machine learning capability that my research group at the University of Cambridge had built up. I wrote before on my soon defunct academic group pages what IoT has to do with systems biology. A lot. In brief, complex biological systems share many properties with complex processes between interconnected technical systems. On the abstraction level required for data science, the methods are identical. For a brief overview of the range of methods and tools we’ve developed, see:

  • Daphne Ezer, Vicki Moignard, Bertie Gottgens, Boris Adryan PLoS Computational Biology 12(8), e1005072 (2016) Determining Physical Mechanisms of Gene Expression Regulation from Single Cell Gene Expression Data
    (Mathematical modelling based on vast amounts of numerical data of unclear distribution)

  • Chie Hosono, Ryo Matsuda, Boris Adryan, Christos Samakovlis Nature Cell Biology 17, 1569–1576 (2015) Transient junction anisotropies orient annular cell polarization in the Drosophila airway tubes
    (Network analysis)

  • Lenka Skalska, Robert Stojnic, Jinghua Li, Bettina Fischer, Gustavo Cerda‐Moya, Hiroshi Sakai, Shahragim Tajbakhsh, Steven Russell, Boris Adryan, Sarah Bray The EMBO Journal 34, 1889-1904 (2015) Chromatin signatures at Notch‐regulated enhancers reveal large‐scale changes in H3K56ac upon activation
    (Markov-model based classification)

  • Jin Hong, Robert Stojnic, Boris Adryan, Anil Ozdemir, Angelique Stathopoulos, Manfred Frasch PLoS Genetics 9(1), e1003195 (2013) Genome-Wide Screens for In Vivo Tinman Binding Sites Identify Cardiac Enhancers with Diverse Functional Architectures
    (Machine learning with a large number of features from small classes)

  • Robert Stojnic, Audrey Fu, Boris Adryan PLoS Computational Biology 8(11), e1002725 (2013) A graphical modelling approach to the dissection of highly correlated transcription factor binding site profiles
    (Learning from highly correlated features)

  • Radu Zabet, Boris Adryan Bioinformatics 28(11), 1517-24 (2012) A comprehensive computational model of facilitated diffusion in prokaryotes.
    (Understanding a system through agent-based simulation)


Over the summer I’ve relocated to Germany, and joined Zühlke Engineering GmbH in Eschborn near Frankfurt. Working with Zühlke is a great opportunity to leverage the exposure of a company that’s established and well-known in the European enterprise IT and industrial hard- and software solution market. What does this mean in practise?

  • First and foremost: I can focus on the conversation between us -as solution providers- and customers who already know that they’re going to get a technically excellent solution. (Establishing that convo takes a lot more legwork if you’re a one-man show, and was just a waste of my time.)
  • My skills are going to see a lot more use in a much wider area than I could have attracted as thingslearn Ltd. For obvious reasons I can’t be more specific, but in the first two weeks at Zühlke I have gained more insight into novel and exciting use cases of IoT analytics than in the previous two years doing thingslearn.

In conclusion: If you, dear reader, are interested in an IoT/M2M solution and/or require consulting in any part of the IoT stack, including analytics, with Zühlke you’ve got a partner with a proven track record. And yes, you can ask for me specifically. That’s an option. You can reach me on boris.adryan AT zuehlke.com.