(Invited) Directed Evolution and Machine Learning for Improved ssDNA-SWCNTs Sensor for Nitric Oxide Detection

Abstract

Single-walled carbon nanotubes (SWCNTs) show great potential for biosensing applications due to their unique optical properties. SWCNTs emit fluorescence in the near-infrared (NIR) region that is stable, biotransparent, and highly sensitive to changes in their environment. The biocompatibility and the selectivity of their interactions can be engineered through SWCNT surface functionalization. One of the most common functionalization approaches is the non-covalent wrapping of SWCNTs with single-stranded DNA (ssDNA). Though the ssDNA sequences can be varied to control the optical response of these SWCNT sensors, the dependence of the optical response on the sequence is unknown and unpredictable. This lack of information on the relationship between the ssDNA sequence and sensor performance is a major bottleneck in engineering SWCNT-based optical sensors.In this work, we develop a guided approach to engineer optical ssDNA-SWCNT sensors. Using a combination of machine learning and directed evolution[1], we designed an optical sensor for nitric oxide (NO), a pro-inflammatory mediator that induces inflammation[2]. We demonstrate a 7-fold enhancement in sensor response compared to the state-of-the-art sensor based on the (AT)15 sequence[3]. This approach thus provides a powerful means of overcoming existing bottlenecks in SWCNT sensor design, ushering a new generation of near-infrared technologies for biomedical research and clinical diagnostics.