Figure 1: Wet Lab and Dry Lab Design
Drylab
Our focus with drylab this year was to discover a solution to design and optimize new PETase sequences such that they have an increased rate of PET degradation. To do this, we aimed to design a machine learning algorithm that utilized a recurrent neural network (RNN) with specific filters, such that the output will display a selected number of PETase sequences, each of which are potential candidates for increased enzymatic activity.
We began the process with a literature review and research with PETase completed in the past. We saw that the iGEM Toronto team attempted this two years ago in 2019, however we did find some potential for improvement upon their pipeline. Following our research, we gathered several sequences found in literature and in other teams’ tests to include in our training dataset. We tested and reworked our pipeline with various training datasets to improve and clean up its output based on the filters created within the pipeline. Once we were able to achieve this, we generated our sequences and sent three of our candidates over to the wetlab team for validation, where they will test the mutants for enzymatic activity.
Wetlab
Our focus with wetlab this year was to carry out wildtype PETase protein purification and ultimately purify and gather data for the mutant PETase sequences that drylab made. To do this, we began early in the summer by doing literature review on the PET degradation process itself, from looking into key enzymes involved as well as researchers in this field that we ask questions to.
Following our research, we began to slowly design protocols from scratch, as seen in our Notebook section. We’ve made countless revisions each time to our protocols based on what we found in research papers and what past iGEM teams did. During our “build” stage, we began working with e.coli and our PET sequences where we kept track of colony growth, color changes in culture media as well as beginning to run tests and assays such as induction and SDS-PAGE.