Cryogenic electron microscopy (cryo-EM) and tomography (cryo-ET) are powerful tools for visualizing the structure of biological macromolecules in vitro and in situ. Current methods can identify conformational ensembles of macromolecules through classification, but these methods require averaging over tens- to hundreds-of-thousands of particles, limiting their application to less abundant cellular complexes. We present MOSAICS (molecular in situ atomic coordinate scanning) as a quantitative strategy that leverages 2D-template matching to directly compare molecular models against cryo-EM images. MOSAICS can identify structural differences between molecular populations in situ using fewer particles than required for classical 3D reconstruction techniques—extending in situ structural analysis beyond only highly abundant cellular complexes. Applying MOSAICS to visualize the protein compositional changes during nuclear 60S ribosome assembly in yeast, we find that the order of ribosome biogenesis factor reorganization differs from in vitro models. MOSAICS is available as an open-source, easily installable Python package alongside Leopard-EM, an extensible Python package for running GPU-accelerated 2D-template matching, and is compatible with other workflows. MOSAICS presents a new paradigm for structural cell biology by enabling spatially resolved, high-resolution structure probing of RNA and protein structures.
Giammar et al. (Sun,) studied this question.