journal menu
![[Figure 1]](bar5002fig1mag.jpg)
|
|
Figure 1
Overview of the CryoLike software pipeline. (a) Workflow diagram: the input images are read from a STAR or CryoSPARC file (containing the metadata) and an MRC file (containing the image data). A non-uniform fast Fourier transform (NUFFT) converts the images from physical space to Fourier space. The sampling density of the images and templates is determined by the resolution factor and viewing-angle distance. A set of templates is derived from an atomic model (from a PDB file) or volume data (from an MRC file). The contrast transfer function (CTF) (corresponding to each image) is applied to the templates during the cross-correlation calculation. The cross-correlation values between the images and the templates at all possible image parameters τ are calculated. The outputs of CryoLike are the optimal cross-correlation, optimal parameters τ that maximize the cross-correlation and the likelihood values for each image. (b) Image transformation: the physical image and template are transformed via NUFFT between physical space represented by Cartesian coordinates and Fourier space represented in polar coordinates. The Fourier polar images undergo a 1D fast Fourier transform (1D FFT) across the polar angular axis, giving the Fourier–Bessel coefficients. The Fourier–Bessel coefficients of an image and a template are multiplied, integrated across the frequencies and an inverse 1D FFT then returns them to the Fourier domain, resulting in the cross-correlation with respect to the rotation angle (bottom right). |
![[Figure 1]](bar5002fig1.jpg)
 | STRUCTURAL BIOLOGY |
ISSN: 2059-7983
