Multimodal cellular context enables molecular phenotyping from single-molecule trajectories

Abstract

Single-molecule tracking (SMT) provides direct measurements of molecular dynamics in living cells, but motion alone rarely specifies the underlying biological process, without accounting for the cellular environment in which that motion occurs. Here, we present MIST (Multimodal Interpretation of Single-molecule Trajectories), a framework that brings the paradigm of image-based profiling to single molecules by integrating trajectory dynamics, emitter properties and cellular context into a common, reusable phenotype space. Using a reference panel of nine proteins in live Escherichia coli, we show that multimodal context more than doubled protein-specific information relative to trajectory features alone, a gap that learned trajectory representations did not close. MIST allowed de novo identification of shared transcription-translation phenotype at the nucleoid periphery. A fixed reference space detected effects of previously uncharacterized MukB ATPase mutations without retraining. MIST provides an interpretable and extensible framework for comparing proteins, mutants and conditions within a common space of molecular phenotypes.