AI and Engineered Yeast Detect PFAS in Municipal Water

Researchers at Columbia University and the City College of New York have developed a novel biohybrid sensor designed to detect per- and polyfluoroalkyl substances, commonly known as PFAS or forever chemicals, in municipal water supplies. The innovation addresses a critical bottleneck in environmental monitoring: current detection methods rely on costly mass spectrometry and commercial laboratories, often resulting in weeks-long delays that hinder timely regulatory compliance and treatment optimization. Under recently implemented EPA regulations, approximately 150,000 public water systems across the United States must frequently test for PFAS. The new device, led by Columbia chemistry and systems biology professor Virginia Cornish alongside civil engineering professor Alex Rosenthal at City College, offers a rapid, low-cost alternative. By integrating engineered biology, artificial intelligence, and semiconductor technology, the system provides real-time concentration readings without laboratory infrastructure. The detection mechanism centers on Saccharomyces cerevisiae, or baker's yeast, genetically modified to function as chemical sensors. Previously, designing synthetic receptors capable of binding specific PFAS molecules required years of manual laboratory work. Leveraging an AI platform developed by Mohammed AlQuraishi's systems biology lab, the research team has compressed the design-build-test cycle to mere weeks. When these AI-optimized receptors bind to target compounds such as PFOA and PFOS, they activate a co-expressed fluorescent protein, causing the yeast to emit a green light proportional to the contaminant concentration. To operationalize the biological sensors, the team has freeze-dried the engineered yeast and embedded them into a porous, paper-like substrate. Readout is handled by complementary metal-oxide-semiconductor, or CMOS, sensor chips originally engineered by Ken Shepard's laboratory at Columbia's School of Engineering and Applied Science for computer-brain interfaces. These optoelectronic chips detect the fluorescent signals and wirelessly transmit data to a handheld reader, delivering instant concentration metrics. Users can simply dip the disposable sensor stick into a water sample and receive immediate results. The biohybrid architecture merges living biological components with solid-state electronics to achieve functionality unattainable through either approach alone. According to team assessments, the per-unit cost of the sensor sticks and readers will be orders of magnitude lower than traditional analytical methods, potentially saving municipal systems hundreds of thousands of dollars annually. More importantly, the technology enables continuous monitoring of water treatment infrastructure, allowing operators to adjust filtration processes dynamically and replace components proactively. Industry professionals have expressed strong interest in the deployment of the system, citing its potential to standardize PFAS surveillance across aging water networks. Beyond environmental applications, the researchers note that the AI-accelerated sensor design framework could establish a new paradigm for yeast-based biosensors in clinical diagnostics and public health. As regulatory pressure intensifies and water quality standards tighten, this integrated biological-electronic platform positions itself as a scalable solution for next-generation contaminant monitoring.