General

Although our project was not fully completed, we still believe that through our modest efforts, we have made a small contribution to the iGEM community. Meanwhile, on this page, we would also like to share our outlook for future work. Specifically, in the Wet Part, we uploaded several gene fragments and plasmid sequences that could serve as resources for future research. We also carried out some cloning experiments, from which we gained valuable experience. In the Dry Part, inspired by our experiments and project objectives, we designed two relatively complete models that helped us gain a deeper understanding of the project. In Human Practices, despite our very limited resources, we still made efforts to learn about the background of our project topic — agricultural mulch films — and to communicate and share our project and its theme with others.

Wet Part

In the Wet Part, we designed a number of parts, including the alkB gene and the alkB gene cluster sequences extracted from Rhodococcus via NCBI (), and uploaded the dual-host plasmid pUWL201, which can function in both E. coli and Streptomyces. Additionally, we designed a system capable of expressing alkB in Streptomyces, with the PermE promoter and fd terminator added upstream and downstream of the alkB gene, respectively. Moreover, we carried out some cloning and conjugation transfer experiments, which may serve as useful references — for example, our procedures for preparing competent cells and culturing Streptomyces. Although we humbly acknowledge that our work is rather modest, we still hope it can be of some value to others.

We have designed a plasmid based on phage phiC31, which can integrate genes directly into the chromosome. In the future, we hope to use this plasmid together with our existing results to complete our measurement experiments.

Dry Part

Our Dry Part mainly consists of two models. The first one is a model we designed to assist in selecting mulch films and evaluate whether a specific type of mulch film is suitable for use. Using a Random Forest Classifier, we developed a model that achieved a final accuracy of 0.83, which can roughly determine what type of mulch film should be used in different regions across China. We also designed an interactive program that allows users to input relevant information so that the model can make its own predictions. (Due to time constraints, we have not yet translated this part into English.)

The second one used the Evo-1 DNA language model and LoRA fine-tuning on Streptomyces CDS data to simulate horizontal gene transfer and optimize the alkB gene from Rhodococcus erythropolis. Through a Metropolis–Hastings MCMC algorithm, we introduced constrained mutations that preserved enzyme activity while improving host compatibility. The optimized sequences were evaluated by CAI, GC content, and protein structure prediction (pLDDT ≈ 94.5, TM-score ≈ 0.98). This work provides a reusable computational framework supporting wet-lab design and demonstrating how machine-learning methods can guide gene adaptation in synthetic biology.

Human Practice

In our Human Practices section, the work we conducted was relatively limited. Nevertheless, we believe that the survey data we collected can still reflect the general situation of mulch film usage in China, as we selected representative cities and regions for our research. Additionally, the mini-presentation we designed may also serve as a useful reference for others. In the long run, our student community may help attract more people to learn about iGEM or even encourage them to begin conducting simple research projects.