Modelling
During the CryptoGErM project we came across several problems and questions that we could not solve in the lab. Therefore we made use of several different modelling approaches. Using artificial intelligence, mathematical models and protein-DNA interaction models we could draw essential conclusions for our project.
Belief-Desire-Intention Model
We asked ourselves what happens if someone tries to hack our system (see photo). How would this person behave? Using an artificial intelligence (AI) model, we explored the behaviour of the people who use, or try to abuse, the CryptoGErM system. The people are modelled in a multi-agent simulation using genetic algorithms. This model helps us to predict the and intentions of the users and intruders of our system.
Random mutations in Bacillus subtilis
We realized a random mutation would be unpleasant (see photo) in our CryptoGErM project. The data stored in the CryptoGErM system might be endangered by random mutations in the DNA. This article contains research of the literature on known mutation rates of B. subtilis under the special conditions and treatments that will be applied.
Random mutations modelling
After exploring the mutation rates in B. subtilis, we asked how will they affected our encrypted message and key in the genome. Therefore in this section, we computationally modelled the probability of random mutation causing our message in the DNA to be corrupted.
Decoding fidelity
Our sequence of the encrypted message is integrated in the genome of B. subtilis, but how sure can we be that we get the same sequence back using common techniques. In the decoding fidelity section we explore how accurate the current sequencing techniques are and how strong our decoy encryption security layer is.
Searching for an optimal strain
We plan to hide our key sequence in one B. subtilis strain, which is mixed with decoy B. subtilis spores. To select the key strain we found the photoswitchable antibiotic spirofloxacin as an interesting option. We modelled the protein-DNA interactions in order to search for an optimal strain for our photoswitchable antibiotic experiments.
