Yeast Gene Edit Prediction for Stable Fermentation Composition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing fermentation processes for producing compounds in yeast are time-consuming and unpredictable, leading to variations in product quality and flavor due to gene editing, requiring repetitive experiments to achieve desired outcomes.
Innovation Solution
A method and system utilizing a tree-based machine learning algorithm and regularized linear model to simulate gene editions in yeast, predicting the impact on gene expression and product composition, followed by validation through high-density expression arrays and chemical analysis to select engineered organisms with desired protein expression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If gene editing is applied to increase yield of target compound, then productivity is improved, but unintended gene expression impact causes manufacturing precision to deteriorate
Solution Approach 1:
The patent applies preliminary action by using computational models and machine learning algorithms to predict gene expression outcomes before actual fermentation experiments are conducted. The system simulates the effects of gene edits on metabolic pathways and predicts unintended consequences, allowing researchers to select optimal gene editing strategies that maximize yield while maintaining product quality, thus avoiding repetitive wet lab experiments.
2Manufacturing precision
If traditional fermentation experimentation is performed to determine quality, then manufacturing precision is improved, but duration of action worsens due to lengthy time-consuming process and repetitions
Solution Approach 1:
The patent employs copying by creating virtual replicas of fermentation processes through computational models. Instead of repeatedly conducting physical wet lab experiments, the system uses machine learning models trained on existing experimental data to simulate and predict outcomes of different gene editing scenarios. This virtual copying allows rapid evaluation of multiple hypotheses without the time cost of repeated physical experiments, significantly reducing the duration needed to determine product quality.
3Productivity
If gene edited cells are used during fermentation, then productivity is improved, but object-generated harmful factors worsen due to unintended gene expression affecting flavor and quality
Solution Approach 1:
The patent implements feedback by using machine learning models that continuously predict gene expression outcomes based on simulated fermentation data. The system incorporates feedback loops where predicted gene expression results are fed back into the model to refine predictions and identify potential harmful unintended effects before they occur in actual fermentation. This allows real-time adjustment of gene editing strategies to eliminate harmful effects while maintaining high productivity.
Data Source
AI summary
A method includes (i) selecting a gene of a yeast to interact with one or more compounds for a mixture; (ii) creating a plurality of gene editions based on the gene; (iii) simulating expression of the plurality of gene editions to determine impact on the gene and a plurality of other genes; (iv) selecting a plurality of designated gene editions that satisfy at least one predetermined criteria; (v) simulating a reaction with the mixture and the yeast having individual ones of the plurality of designated gene editions; (vi) determining an expected final composition after a plurality of the simulated reactions; (vii) correlating data on a plurality of attributes of the expected final compositions to the desired final composition profile; and (viii) selecting and providing one or more expected final compositions and related gene edition data that most closely match the desired composition profile.