On the explicit use of experimental images in high resolution cryo-EM refinement.
Jacqueline CherfilsJorge NavazaPublished in: F1000Research (2019)
Single particle cryogenic electron microscopy (cryo-EM) is transforming structural biology by enabling the analysis of difficult macromolecular specimens, such as membrane proteins or large complexes with flexible elements, at near atomic resolution with an accuracy close to that of X-ray crystallography. As the technique continues to improve, it is important to assess and exploit its full potential to produce the most possible reliable atomic models. Here we propose to use the experimental images as the data for refinement and validation, instead of the reconstructed maps as currently used. This procedure, which is in spirit quite similar to that used in X-ray crystallography where the data include experimental phases, should contribute to improve the quality of the cryo-EM atomic models.
Keyphrases
- electron microscopy
- high resolution
- deep learning
- electronic health record
- convolutional neural network
- big data
- optical coherence tomography
- mass spectrometry
- magnetic resonance imaging
- risk assessment
- artificial intelligence
- machine learning
- human health
- data analysis
- fine needle aspiration
- tandem mass spectrometry
- ultrasound guided