Synthetic Dataset Creation » History » Revision 8
Revision 7 (Dmitry Lyumkis, 06/07/2010 11:20 AM) → Revision 8/18 (Dmitry Lyumkis, 06/07/2010 11:29 AM)
h1. Synthetic Dataset Creation
This method uses projections of a 3D model in order to create a synthetic dataset. Although it can be modified according to the options specified, the scheme consists of 12 basic steps, as shown and summarized below:
!http://emg.nysbc.org/attachments/294/Supplementary_Figure_5_-_Synthetic_dataset_creation.png!
# a model is chosen from which projectinos are created
# the model is projected either in an even distribution or with axial preference
# the projections are randomly rotated in the XY plane
# the projections are randomly shifted in the XY plane
# white Gaussian noise is added
# a contrast transfer function is added according to the specified defocus parameter and the spherical aberration constant of the microscope
# an envelope function is added according to an experimentally determined decay function from 3000 real micrographs (see Voss, N, Lyumkis, D. et al, JSB (2010) 169, 3, 389-98).
# a second level of Gaussian noise is added, usually to bring the signal-to-noise ratio down to ~0.05, consistent with real ice data (see Baxter, WT, et al, JSB (2009) 166, 2, 126-32).
# (optional) the CTF is estimated by ACE2 and corrected
# (optional) the particle is band-pass filtered
# final particle of 50S ribosomal subunit with a SNR of 0.05
# particle is added to a growing stack of synthetic particles