Sep 30, 2023
Image Processing and 3D Reconstruction
This protocol is a draft, published without a DOI.
- 1University of California, Berkeley;
- 2Aligning Science Across Parkinson's
Protocol Citation: Annan SI Cook 2023. Image Processing and 3D Reconstruction. protocols.io https://protocols.io/view/image-processing-and-3d-reconstruction-c2n9ydh6
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: September 22, 2023
Last Modified: May 31, 2024
Protocol Integer ID: 88513
Keywords: ASAPCRN
Funders Acknowledgement:
Aligning Science Across Parkinson's
Grant ID: ASAP-000350
Abstract
This workflow was used to analyze a Krios dataset of the PI3KC3-C1/RAB1A Complex and generate a reconstruction of three distinct conformational states of the VPS34 lipid kinase domain.
Attachments
852-2197.pdf
37KB
Materials
Materials and Software
- cryoSPARC v3 software
- UCSF ChimeraX v1.5 or similar software
- High-performance computing cluster or powerful workstation for computational tasks
Data Import
Data Import
Import the raw Cryo-EM data sets into cryoSPARC v3.
Motion Correction and Fourier Cropping
Motion Correction and Fourier Cropping
Apply motion correction to the super-resolution movies.
Perform Fourier cropping 2x on the motion-corrected data using cryoSPARC's implementation of Patch Motion Correction.
CTF Determination
CTF Determination
Use cryoSPARC's Patch CTF Estimation for Contrast Transfer Function (CTF) determination.
Particle Picking and Training a Model
Particle Picking and Training a Model
Manually pick single particles from a selection of micrographs covering a range of defocus values.
Train a particle-picking model using Topaz.
Particle Extraction and Binning
Particle Extraction and Binning
Extract particles with an appropriate box size (e.g., 400x400x400 or ~1.5x the diameter of a single PI3K complex) to ensure retention of delocalized CTF information.
Optionally, bin the extracted particles 4x to increase computational speed for subsequent processing.
Initial 2D Classification
Initial 2D Classification
Apply two-dimensional (2D) classification to the extracted particles.
Exclude obvious junk particles from further processing based on the 2D classification results.
Heterogeneous Refinement
Heterogeneous Refinement
Use a junk class from the early rounds of an ab initio run, along with a map generated using an apo PI3K model in UCSF Chimera using molmap at a resolution of 20 Å, for heterogeneous refinement.
Iterate this process for 3-4 rounds until a healthy substack of particles is evident by 2D classification.
Ab Initio Reconstruction and Particle Cleanup
Ab Initio Reconstruction and Particle Cleanup
Perform a three-class ab initio reconstruction for a final particle cleanup.
Should result in a clearly clean class of particles.
High-Resolution Refinement
High-Resolution Refinement
Re-extract the particles at a full 400-pixel box size.
Perform homogeneous refinement using the ab-initio model and the clean particle stack to generate a high-resolution model and particle alignments for downstream classification.
3D Classification without Alignment
3D Classification without Alignment
Conduct 3D classification without alignment using a large mask on the putative kinase domain and create 50 classes.
Selection of Classes for Further Analysis
Selection of Classes for Further Analysis
Select the three most populated classes from the group of 50.
Heterogeneous Refinement of Selected Classes
Heterogeneous Refinement of Selected Classes
Perform heterogeneous refinement on the selected classes.
3D-Variability Analysis
3D-Variability Analysis
Perform 3D-variability analysis in cryoSPARC in cluster mode for each population showing strong density for the kinase domain.
Using the clusters containing the strongest density, perform non-uniform refinements on each.
Local Refinement
Local Refinement
Conduct local refinement on three parts of the complex: VPS15 pseudokinase domain, RAB1A interface region, and BECN1/ATG14 BARA dimer domain.
For VPS34 Kinase containing classes - perform local refinement with a mask aligned to the particular pose of the kinase domain.
Create masks for these regions using UCSF ChimeraX Volume Tools.
Combining Refined Maps
Combining Refined Maps
Combine the locally refined maps, which correspond to distinct kinase conformations, using UCSF ChimeraX with the `vop maximum` command.
Model Building and Visualization
Model Building and Visualization
Utilize the combined maps for model building and visualization of the structures.