journal menu
![[Figure 1]](jb5069fig1mag.jpg)
|
|
Figure 1
Optimized cryoSPARC workflow using the combined cryo-EM data set. Two cryo-EM data sets were collected from DSP-cross-linked membrane-protein samples. The first data set (3908 movies) was prepared with 0.02% DDM and 0.4%(w/v) CHAPS as an additive, and the second data set (7506 movies) with 0.05% DDM and 0.005%(w/v) cholesterol hemisuccinate (CHS). Motion correction and CTF estimation were performed separately for each data set. 10 141 micrographs were used for particle picking, resulting in 7.13 million particles (estimated diameter 100–200 Å). Two rounds of 2D classification (200 classes, 256 Å mask, 40 online EM iterations, batch size 400) retained 74 668 particles (scale bar 37.8 nm). These classes were used as templates for template-based picking, which yielded 6.3 million particles. After re-extraction and duplicate removal, a final set of 10.13 million unique particles was obtained. 2D classification yielded 1.26 million particles distributed across the best 20 classes. For ArnA, 123 593 particles corresponding to ArnA 2D classes (scale bar 37.8 nm for 2D class averages) were selected and processed through ab initio reconstruction with C1 symmetry, followed by heterogeneous and non-uniform refinement using D3 symmetry. For AcrB, 104 054 particles corresponding to AcrB 2D classes (scale bar 37.8 nm for 2D class averages) were selected and refined using C3 symmetry. Global CTF refinement and DeepEMhancer post-processing were applied to both maps. Handedness correction for AcrB was performed using the Volume Tools utility in cryoSPARC. Particles corresponding to GroEL, cytochrome bo3 oxidase and other minor components were also detected during classification (scale bar 37.8 nm for 2D class averages). |
![[Figure 1]](jb5069fig1.jpg)
 | STRUCTURAL BIOLOGY |
ISSN: 2059-7983
Open 
access
access
