Applications of Machine Learning Techniques in Genetic Circuit Design

Construction of mathematical models to investigate genetic circuit design is a powerful technique in synthetic biology with real-world applications in biomanufacturing and biosensing. The challenge of building such models is to accurately discover flow of information in simple as well as complex biological systems. However, building synthetic biological models is often a time-consuming process with relatively low prediction accuracy for highly complex genetic circuits. The primary goal of this study was to investigate the utility of various machine learning (ML) techniques to accurately construct mathematical models for predicting gene expressions in genetic circuit designs. Specifically, classification and regressions models were built using Random Forrest (RF), Support Vector Machines (SVM), and Artificial Neural Networks (ANN). The obtained accuracy of the regression model using RF and ANN yielded R2 scores of 0.97 and 0.95, respectively, compared to the best score of 0.63 obtained in an earlier study. Furthermore, a classifier model was built using the green fluorescent protein (GFP) measurements obtained from the experiments conducted in this work. Biologists use GFP as an indicator of gene expression, enabling easy measurement of its protein level in the living cells. The measured GFP values were predicted with 100% accuracy by both RF and ANN classifier models while identifying various synthetic gene circuit patterns. The paper also highlights importance of the relevant data preparation techniques to ensure high accuracy is obtained by the utilized ML models.