Tutorial Videos
CryoSPARC tutorial videos.
Last updated
CryoSPARC tutorial videos.
Last updated
For written tutorials, please see:
This playlist contains six recordings covering single particle cryo-EM data processing in CryoSPARC from the 2024 Single-Particle Cryo-EM Image Processing Workshop held by the Stanford-SLAC CryoEM Center (S2C2) in 2024. The original recordings are courtesy of S2C2.
In this opening video, the CryoSPARC Team covers fundamental concepts of cryo-EM, including the basics of data collection using an electron microscope, defocus and the contrast transfer function (CTF), motion correction, picking and extraction of individual single particles, expectation maximization and particle classification in 2D and 3D. Be sure to watch the explanation of what a “Think-Pair” question is at 4:45 - they show up throughout the rest of the videos!
In the first section, Data Collection and Preprocessing, we cover what happens to a cryo-EM sample when it’s in the microscope and how this affects the captured images. We also cover the basics of how image aberrations are typically corrected for in cryo-EM.
Next, we discuss how individual particles are picked and extracted from images. If you’ve ever wondered how you should pick a box size, or what the Nyquist Resolution is, this is the section for you!
We then move on to discuss 2D Classification. We also cover the Expectation Maximization algorithm in this section. If this is your first exposure to cryo-EM, or if you’ve always wondered how high-quality 3D maps are actually computed from very noisy images, this is where you can find that information. We also highly recommend all users to watch the section covering the concept of a reference and pose, starting at 40:23. These are critical concepts used in most of the algorithms for single particle cryo-EM analysis.
Finally, we briefly touch on 3D techniques before moving on to discuss validation.
In this video, we present a step-by-step explanation of one possible way to process micrographs of TRPV1 from EMPIAR 10059. The jobs presented here define what we call a “standard workflow”, which we expect to produce a workable consensus refinement for most single particle cryo-EM datasets. By no means is this the only (or the best) way to process data! It is merely meant to guide new users through the most common set of jobs, and present ways to handle challenges as they arise.
Our standard workflow comprises preprocessing, blob picking, particle curation, template picking, more particle curation, and finally a consensus refinement. We cover each of these steps in detail in the relevant chapters, explaining parameter choices along the way. Additionally, we present some jobs which do not produce the desired result and discuss why they failed.
In this video, we follow the standard workflow established in the previous video to produce a map of a related ion channel, TRPV5 (EMPIAR 10256). However, each TRPV5 particle binds one calmodulin in one of the four C4 symmetric positions. The C4 symmetric map therefore has artifacts due to misaligned calmodulin molecules.
We compare and contrast three techniques for handling this pseudosymmetry in CryoSPARC. First, we try simply re-refining the particles using a global refinement without imposing C4 symmetry. Next, we try classifying the particles with Heterogeneous Refinement or 3D Classification. Finally, we take advantage of the fact that we know the order of the pseudosymmetry and use Symmetry Relaxation.
After working through the TRPV5 case, we return to TRPV1 and resolve the symmetry mismatch between the C4 symmetric channel and the C2 symmetric DkTx linkers. We try the three methods above again with this new target and compare their performance to the calmodulin case.
Unfortunately, during recording, audio was lost for approximately two minutes from 1:08:05 to 1:10:01. We apologize for the inconvenience.
In this video, we investigate Encapsulated Ferritin (EncFer, EMPIAR 10716). Each particle in this cryo-EM dataset has two parts: an icosahedral encapsulin and four D5 symmetric EncFer proteins. The EncFer are contained inside the encapsulin protein, meaning there is a symmetry mismatch between the outer shell and the inner parts. The majority of the case study focuses on improving the resolution of the internal EncFer molecules.
Interestingly, although four EncFer molecules appear tetrahedral, their internal D5 symmetry means that the overall arrangement has no point-group symmetry. This requires the use of two new workflows in CryoSPARC, which we call Group Re-alignment and Custom Symmetry Expansion.
A written companion to this video is available in the CryoSPARC Guide: Case Study: End-to-end processing of encapsulated ferritin.
In this video, we work on a combined cryo-EM dataset of apo and ligand-bound FaNaC1 (EMPIAR 11631 and 11632). Here we present workflows which treat these structures as distinct, discrete states, and discuss important considerations when using jobs like 3D Classification and Heterogeneous Refinement to separate them.
A major point of discussion is the impact of the input poses (from a consensus refinement) on the results of 3D Classification, and how the maps from 3D Classification can be of deceptively poor quality before they are re-refined. During the Q&A session, we also discuss our opinions on how to decide whether a single particle cryo-EM data processing journey is “complete”, and how this decision depends in large part on the goal of the investigator.
In the final video in this cryo-EM data processing series, we consider the same FaNaC1 dataset as in Part 5, but consider the apo and ligand-bound states as existing on a continuous spectrum. We begin by considering the difficulty inherent in extracting the conformation of individual particle images, and why results from these techniques should always be validated using orthogonal, biochemical methods.
We then discuss the theory and practice of 3D Variability Analysis in CryoSPARC. In 3D Variability Analysis, each particle is modeled as a consensus refinement plus some linear combination of difference volumes. This makes the technique computationally lighter-weight than alternatives and can produce good results in many situations.
Finally, we discuss 3D Flexible Refinement (3DFlex). This technique directly models continuous deformation of the consensus refinement using nonlinear, machine-learning techniques. We discuss important steps in setting up 3DFlex, including mesh design and important parameter choices.
These videos from the 2020 S2C2 workshop cover the v3 interface, but go into more detail about CryoSPARC itself, including how projects, jobs, parameters, and inputs/outputs work in the software. Additionally, these videos investigate two additional datasets: the HA Trimer (EMPIAR 10097), NaV 1.7 channels (EMPIAR 10261), and the cannabinoid receptor 1-G protein complex (EMPIAR 10288).
Processing EMPIAR-10288 using CryoSPARC Live.
For a detailed guide on CryoSPARC Live, see:
How to use Reference Based Motion Correction in CryoSPARC v4.4+.
How to construct an automated pipeline of jobs in v4.4+.
Creating a mask in ChimeraX using three different techniques: volume segmentation, volume eraser and molmap.
When might your data benefit from a custom mesh, and the process of mesh creation.
