This program leverages advances in cryogenic transmission electron microscopy (cryo-TEM) and four-dimensional scanning transmission electron microscopy (4D-STEM) to validate by atomic-scale images the locations of atoms in soft matter and confirm the prediction of simulations. Imaging soft matter is challenging because the atoms are not located at precise locations and they are extremely beam sensitive. Using machine-learning-based algorithms reveal the heterogeneous crystal structure of nanosheets formed by polypeptoids using low-dose images. Atomistic molecular dynamics is used to both interpret the images and determine the underlying interactions that govern structure formation.