High-speed Raman2RNA: nondestructive prediction of single-cell expression profiles with total internal reflection Raman microscopy

Single-cell RNA-seq and other profiling assays have provided unprecedented insight into cellular properties, regulation, dynamics, and function. However, because these assays are inherently destructive, they do not allow us to track temporal changes in living cells. To overcome this limitation, we previously developed Raman2RNA (R2R), an experimental and computational framework that employs Raman microscopy and machine learning to infer single-cell expression profiles in live cells. Although this method works well in cell culture, however, the intrinsically weak Raman signal made it challenging to apply to tissue-scale imaging. To this end, we introduce total internal reflection Raman microscopy (TIRR). TIRR simultaneously captures Raman photons along a single line focus without requiring high-power lasers, while still delivering sufficient energy per illumination volume. We demonstrate its advantages across various cell types and tissues, showing that TIRR offers a marked speed improvement over confocal or line-illuminated Raman microscopes. This enhanced, “high-speed Raman2RNA” approach opens the door to broad applications for studying expression dynamics at scale, in vitro and in vivo.