Figure 2
Computational pipeline for extracting domain-specific heterogeneity from cryo-EM 2D particles. 525489 2D particles with estimated contrast-transferring functions (CTFs) and poses from CryoSPARC (Punjani et al., 2017 ) were projected to a low-dimensional latent space by principal component analysis (PCA) using RECOVAR (Gilles & Singer, 2025 ), from which 13 main volumes were found via k-means clustering. Volumes were constructed from them with RECOVAR, showing different up/down states of RBDs. Two volumes, corresponding to the centroids crossed out in the latent space, were excluded from the analysis since they were incomplete particles. |