Data diversity is critical to success when training deep learning models. Medical imaging data sets are often imbalanced as pathologic findings are generally rare, which introduces significant challenges when training deep learning models. In this work, we propose a method to generate synthetic abnormal MRI images with brain tumors by training a generative adversarial network using two publicly available data sets of brain MRI. We demonstrate two unique benefits that the synthetic images provide. First, we illustrate improved performance on tumor segmentation by leveraging the synthetic images as a form of data augmentation. Second, we demonstrate the value of generative models as an anonymization tool, achieving comparable tumor segmentation results when trained on the synthetic data versus when trained on real subject data. Together, these results offer a potential solution to two of the largest challenges facing machine learning in medical imaging, namely the small incidence of pathological findings, and the restrictions around sharing of patient data.

Genetic regulatory proteins inducible by small molecules are useful synthetic biology tools as sensors and switches. Bacterial allosteric transcription factors (aTFs) are a major class of regulatory proteins, but few aTFs have been redesigned to respond to new effectors beyond natural aTF-inducer pairs. Altering inducer specificity in these proteins is difficult because substitutions that affect inducer binding may also disrupt allostery. We engineered an aTF, the Escherichia coli lac repressor, LacI, to respond to one of four new inducer molecules: fucose, gentiobiose, lactitol or sucralose. Using computational protein design, single-residue saturation mutagenesis or random mutagenesis, along with multiplex assembly, we identified new variants comparable in specificity and induction to wild-type LacI with its inducer, isopropyl β-D-1-thiogalactopyranoside (IPTG). The ability to create designer aTFs will enable applications including dynamic control of cell metabolism, cell biology and synthetic gene circuits.

Engineering biosynthetic pathways for chemical production requires extensive optimization of the host cellular metabolic machinery. Because it is challenging to specify a priori an optimal design, metabolic engineers often need to construct and evaluate a large number of variants of the pathway. We report a general strategy that combines targeted genome-wide mutagenesis to generate pathway variants with evolution to enrich for rare high producers. We convert the intracellular presence of the target chemical into a fitness advantage for the cell by using a sensor domain responsive to the chemical to control a reporter gene necessary for survival under selective conditions. Because artificial selection tends to amplify unproductive cheaters, we devised a negative selection scheme to eliminate cheaters while preserving library diversity. This scheme allows us to perform multiple rounds of evolution (addressing ∼10 9 cells per round) with minimal carryover of cheaters after each round. Based on candidate genes identified by flux balance analysis, we used targeted genome-wide mutagenesis to vary the expression of pathway genes involved in the production of naringenin and glucaric acid. Through up to four rounds of evolution, we increased production of naringenin and glucaric acid by 36-and 22-fold, respectively. Naringenin production (61 mg/L) from glucose was more than double the previous highest titer reported. Wholegenome sequencing of evolved strains revealed additional untargeted mutations that likely benefit production, suggesting new routes for optimization.evolution | metabolic engineering | synthetic biology | sensors | biosynthetic pathways

Characterization and standardization of inducible transcriptional regulators has transformed how scientists approach biology by allowing precise and tunable control of gene expression. Despite their utility, only a handful of well-characterized regulators exist, limiting the complexity of engineered biological systems. We apply a characterization pipeline to four genetically encoded sensors that respond to acrylate, glucarate, erythromycin and naringenin. We evaluate how the concentration of the inducing chemical relates to protein expression, how the extent of induction affects protein expression kinetics, and how the activation behavior of single cells relates to ensemble measurements. We show that activation of each sensor is orthogonal to the other sensors, and to other common inducible systems. We demonstrate independent control of three fluorescent proteins in a single cell, chemically defining eight unique transcriptional states. To demonstrate biosensor utility in metabolic engineering, we apply the glucarate biosensor to monitor product formation in a heterologous glucarate biosynthesis pathway and identify superior enzyme variants. Doubling the number of well-characterized inducible systems makes more complex synthetic biological circuits accessible. Characterizing sensors that transduce the intracellular concentration of valuable metabolites into fluorescent readouts enables high-throughput screening of biological catalysts and alleviates the primary bottleneck of the metabolic engineering design-build-test cycle.