Machine learning tools for assisting the development of robust artificial microbial consortia for the degradation of polyolefins
Project leader: Marcus Liwicki, Machine Learning, LTU
Project team: Anjali Purohit, Christakopoulos Paul, Leonidas Matsakas, Ulrika Rova
Duration: 2024-2026
Funded by Kempestiftelserna
Microbes typically exist in complex communities (consortia) that communicate through quorum sensing and share metabolites and enzymes. Recent advances, such as high-throughput screening and sequencing, as well as computational and machine learning (ML) advances has allowed for the design, building, and analysis of complex consortia. This has led to an emerging field in synthetic biology and biotechnology with the aim of designing artificial communities that are as robust, efficient, and flexible as the ones found nature. The aim of this project is to create novel ML approaches to support the engineering of artificial microbial consortia. Recent self-supervised learning and graph neural network (GNN) approaches will be used to create novel self-supervised pretrained GNN as backbone knowledge extractor to finally train generative deep learning methods such as variational autoencoders to generate candidates for new artificial microbial consortia with desirable properties.
Updated: