Abstract
Principal component analysis (PCA) has been widely proposed to analyze flexibility and heterogeneity in cryo-electron microscopy (cryoEM). In this paper, it is argued that (i) PCA is an excellent technique to describe continuous flexibility at low resolution (but not so much at high resolution) and (ii) PCA components should be analyzed in a concerted manner (and not independently).
Highlights
Biological macromolecules can be regarded as flexible objects whose movements, which are continuous in a general case, allow them to perform their physiological functions
We argue that (i) Principal component analysis (PCA) is an excellent technique to describe continuous flexibility at low resolution and (ii) PCA components should be analyzed in a concerted manner
In cryo-electron microscopy (cryoEM), it is customary to interpret each principal component independently. [For instance, the first principal component may represent a rotation of a certain part of the macromolecule, the second principal component a shift of another part etc.; see Figs. 2, 4 and 5 in Punjani & Fleet (2020); the reader should note that this is not a criticism of that specific work, as its authors are fully aware of the importance of the distribution of images in the PCA subspace
Summary
Biological macromolecules can be regarded as flexible objects whose movements, which are continuous in a general case, allow them to perform their physiological functions. Principal component analysis (PCA) has been widely proposed to analyze flexibility and heterogeneity in cryo-electron microscopy (cryoEM) (Tagare et al, 2015; Haselbach et al, 2018; Punjani & Fleet, 2020). This technique assumes that the different conformations that are present in a sample can be constructed as a weighted sum of the eigenvectors of the covariance matrix of the volume. We argue that (i) PCA is an excellent technique to describe continuous flexibility at low resolution (but not so much at high resolution) and (ii) PCA components should be analyzed in a concerted manner (and not independently)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have