Basics of single-particle cryo-EM image processing

My virtual lecture at the Midlands Regional Cryo-EM Workshop on the introduction to single-particle analysis. The lecture is partially based on work performed at MRC Laboratory of Molecular Biology (laboratory of Paula da Fonseca) and Leiden University (laboratory of Marin van Heel).

Midlands Regional CryoEM Facility: https://midlandsinnovation.org.uk/Cryo

Timestamps:
0:00:00 Introduction and the goal of the lecture
0:00:53 Background on the 20S proteasome example
0:02:00 Gabe Lander’s film on the cryo-EM pipeline
0:05:20 Why did the “Resolution revolution” happen?
0:06:29 Limitations of cryo-EM: why is it still not a routine technique
0:08:03 Outline of the cryo-EM experiment
0:10:50 Pipeline of the single-particle image processing
0:11:58 What is denoising?
0:13:28 Timeline of single-particle software developments: from Spider and Imagic to Relion and Cryosparc
0:16:06 Movie-mode data collection: why do we use it?
0:17:02 Was movie-mode data collection ahead of time? The original paper by Kunath et al., 1984 suggested the algorithm implemented in modern software packages
0:19:03 Why is it important to do manual data screening?
0:20:23 Understanding your data in reciprocal (Fourier) space
0:21:56 Low-pass and high-pass filters
0:23:06 CTF-determination
0:24:06 What do power spectra look like with and without movie alignments?
0:24:31 Selection of micrographs based on the power spectrum
0:25:32 How would a typical real-life micrograph look like with a low defocus?
0:27:59 Particle picking was done by PhD students manually in the old days
0:28:35 How automatic particle picking is implemented
0:29:36 Reference bias in particle picking: “Einstein from noise” experiment
0:31:59 The “Mao case”: what went wrong
0:36:09 Particle extraction: pre-processing, normalisation, centring and binning
0:38:28 Denoising by averaging. Why do we need to do alignments?
0:39:24 The idea of alignments and 2D-classification
0:42:59 Going from 2D to 3D. The idea of 3D-reconstruction
0:45:10 What is projection matching?
0:47:43 How to get an initial 3D-reconstruction?
0:49:16 Refinements and validation. Fourier Shell Correlation
0:53:06 Illustration of the evolution of 3D-refinements
0:53:48 Estimation of local resolution by Resmap, which might be lying to you
0:54:04 Illustration of the importance of evaluation of the final results. The idea of 3D classification.
0:57:37 What is done after 3D-classification?
0:58:07 The idea of building a pipeline using various software packages
0:58:46 Why should one be careful with on-the-fly data processing?
0:59:14 Take-home messages