Many potential mutagenic substances exist in the environment, and they may cause damage to
DNA in chemical or physical ways. Accumulating mutagenic harm may lead to gene mutations, causing cell
damage, death, or transformation. Therefore, this year, our team constructed a system that
quantitatively analyzed mutagenicity by combining the gene circuit design with the AI model. For all
chemicals that already have known structures, we used the AI system to analyze its mutagenicity
quantitatively. We built up our powerful AI with the results we found after paper research indicating
that substructures contribute to mutation.
Moreover, we used a chemical mutagenicity database and support vector machine algorithm to
train the AI. It enabled us to give out precise mutagenicity with users inputting chemical names.
Concerning a mutagenic factor with an unknown structure or an unidentified target, for example, UV
light, radiation, and complex chemical compounds, by using our designed gene circuit and model, we could
also perform qualitative and quantitative analysis. What we have done, essentially, was to transform
toxin genes into E. coli, respectively, and observe its growth with O.D. measurement. According
to the
designed gene circuit, when the transformed E. coli faced a situation with mutagen, the growth
would
show its specific growing process. After documenting and gathering the results from different situations
and model simulations, we could thus put the data into qualitative and quantitative analysis.
For chemicals that already have known structures, a Mutagenicity Prediction AI based on relationships between chemical structures and mutagenicity was developed after paper surveying. Moreover, we collected specify chemical structures as training data and the result of Ames test as target data for machine learning by Support Vector Machine algorithm (SVM). Enabling the public to learn mutagenicity with a few clicks online with our user-friendly user interface.