Case Study: DTU Biosustain achieves 106% higher GFP synthesis and up to 74% higher tryptophan titer with TeselaGen

Using Machine Learning for Opt imizat ion of Cellular Factories To Produce Indust rial Products

DTU Biosustain’s Jensen Lab (Synthetic Biology Tools for Yeast) needed an efficient way to train machine learning models to predict genotype-to-phenotype relationships and optimize metabolic pathway designs for tryptophan production in yeast. They partnered with TeselaGen and used TeselaGen’s DISCOVER module — its ML tools and compute infrastructure — to combine mechanistic models with machine learning and move beyond labor-intensive trial-and-error engineering.

TeselaGen implemented its DISCOVER ML pipeline to design, train, and execute predictive models on the lab’s genetic construct data, generating diverse candidate genes and cell-factory designs. The collaboration produced large gains: a GFP synthesis rate 106% higher than the already improved platform, and improvements in tryptophan titer and productivity of 74% and 43%, respectively, compared with the best designs used for training.

DTU Biosustain

Michael Krogh Jensen

DTU Biosustain