Notes on using SAVR 1. When boxing out the particles, choose sizes that are "friendly" to FFT. Allow 10-20% extra padding space for FFT. Most FFTs are based on the Cooley-Tukey algorithm and have a speed advantage when the size is a multiple of small prime numbers (friendly). 2. The mapsize of the final reconstruction does not have to be the same size as the boxed particles. In fact, is should be odd (for example, 699). A size slightly larger than the particle should be sufficient. This willl specify what to reconstruct, and padding should not be reconstructed. However, both the reconstruction and the particles will have the same scale. At this time, the map should be <700 pixels due to program/memory limitations. The reconstructed maps are hemispheres. 3. The starting model should be in the same scale as the particles, but it may be a half map. Padding is not required. It is required that the map be in the MRC convention (5-fold along the Z-axis and 2-fold along the Y-axis). Specify whether the model has been CTF-corrected under the 'SMCTFcorrect' flag: 0 = the model is not corrected (protein is black in Hong's convention); 1 = the model is CTF-corrected. 4. MRCRefine specifies the resolution to refine to at each iterative step. The default spacings are linear in reciprocal space. 5. RefineMore specifies how much further to refine, starting at the last cycle specified by MRCRefine. Otherwise, changing ResRange will change the spacings in MRCRefine, so that the whole process is repeated. 6. ResRange and SMResolution specify the resolution of the reconstruction and the starting model, respectively. ResRange is the low and high resolution limit of this reconstruction (see MRCRefine). In order to reconstruct the low-resolution map, lower the starting model's resolution. 7. In the micrograph section, the required parameters are b-factor, defocus, and image file (eg *.img and *.hed). A/pix may be specified in the header section. 8. PostiveContrast: This flag specifies if the images have the protein as white (1) or black (0). If black, the program will invert the density before processing. 9. ICFactor 10. IMaskSlope 11. The ".cshrc" file is required. If the environment is specified in another shell, then the other file can be used with "source": % cat .cshrc source /home/juan/.tcshrc 12. Do not source from "tcsh" to "zsh" to run virus.py. This may cause errors due to environment variables/paths that are not inherited. 13. If using Z-shell, make sure both .zshrc and .zshenv are present. 14. When processing on different architectures, be aware of little/big Endians. 15. Plan to have 4X the diskspace required to store all the particles. 16. The *.img and *.hed files must be writable: % ls -l -rw-r--r-- 1 juan user 83968 Jun 27 19:07 jj0290.hed -rw-r--r-- 1 juan user 1700035200 Jun 27 19:07 jj0290.img -rw-r--r-- 1 juan user 83968 Jun 27 19:07 jj0291.hed -rw-r--r-- 1 juan user 1700035200 Jun 27 19:07 jj0291.img 17. CCMLPLimits specify the phase residual range for processing each particle. The program will start with a low phase residual and linearly increase the value. At low resolution, the signal-to-noise ratio is high, so the PR is more stringent. At high resolution, SNR is low, so it is relaxed. 18. CCMOLimits/CCMPLimits: If the orientation (degrees) or the centering (pixels) of the particle exceeds this value, the particle will be eliminated from the reconstruction, but will be refined again in the next iteration. 19. Environment variables, paths, files may not be accessible from all the nodes in a cluster. For example, /usr/lib/libtiff.so.3 was not accessible, so it was copied into the local EMAN/lib directory. 20. Use Python version 2.1 or newer. 21. On the clusters, the files <.eman/mparm> and are required. This is to prevent the processes from running on the root. These files will specify the nodes that are available, and they must be synchronized. There should be no blank lines in these files. See example: ssh <# cpus per node> 1 22. If a process the sleeping for a long time, it is probably dead. Kill the job and the process from every node before restarting. 23. Check "lg2map.log" and "emicolg.log" in "PRoot/mrc-2d-dir/" for any errors during the reconstruction. Also pipe the output from the processing to a file for subsequent debugging. For example: % screen -S hsv-reconstruction % virus.py hsv.prj >& log.01 Ctrl-A-D % screen -ls There is a screen on: 1234.hsv-reconstruction (Detached) 1 socket in /home/juan/tmp. % cat log.01 The 3 refinement cycles are: [30.0, 25.2, 21.6] Starting cycle 0 24. The file PRoot/MRC-CTF-DB.dat contains the CTF parameters for each micrograph. 25. To restart from a certain resolution cycle, delete the maps (*.mrc) and orientation files (*.dat) from PRoot/mrc-3d-dir and the projections (*.tnf) from PRoot/mrc-proj-dir General reconstruction procedure and checkpoints: 1. The files for the initial model are save in 'PRoot/initmodel-dir': 100.cen . center of the particle 100.log . initial orientations using 3 criteria template.dat . self-common lines orientation of 3 good raw particles . used to eliminate bad particles . checkpoint globalRefined.dat . cross-common lines between all raw particles for orientation search . used for building initial model . checkpoint The micrographs are ranked by how close their first peak is to 30A. Those nearest 30A are used for building the initial model. 2. Projections from the initial model are used in a global search for the orientation of all the particles. This uses the projection-matching from the program 'WaveOrt'. Results from the particles of each micrograph are saved in their respective *.hed files. In the 'PRoot' directory: wave-proj-dir/projections-for-cycle-of-30.00Angstrom-resolution.img wave-proj-dir/projections-for-cycle-of-30.00Angstrom-resolution.hed . hundreds of projections from the initial model . used for projection-matching to roughly determine the orientation of the raw particles jj0198/eman-workdir/jj0198.hed . the orientation of the particles from jj0198.img are saved here jj0198/eman-workdir/Cycle-of-30Angstroms-is-done . checkpoint 3. Cross-common lines are used to in the local search for the particle orientations. In the 'PRoot' directory: mrc-proj-dir . directory containing the projections from the initial model jj0198/mrc-workdir/Wave-ort-cen-cycle-of-30.00Angstrom.dat . global orientations of the particles in this micrograph (from projection-matching) . input to cross-common lines orientation search . checkpoint jj0198/mrc-workdir/jj0198-refined-to-res-30.00Angstrom.dat . results from the cross-common lines local orientation search (see below) . used for the 30A reconstruction cycle . checkpoint 4. When the orientations have been determined for all the particles, they are pooled together in order to reconstruct the map. The particles are divided into 2 groups to compute the Fourier shell correlation, but the map is reconstructed using all the particles. This map will be used in the next cycle (starting at step 2) as the initial model. If specified, an email will be sent after each cycle. mrc-3d-dir/All-particles-refined-to-30.00Angstrom-ortcen.dat . Orientations of all the particles mrc-3d-dir/icos5f-refined-to-30.00Angstrom.mrc . Map built from all the particles . checkpoint mrc-3d-dir/All-particles-refined-to-30.00Angstrom-ortcen-set-0-of-2.dat . Orientations of the particles in one set mrc-3d-dir/icos5f-refined-to-30.00Angstrom-set-0-of-2.mrc . Map built from the particles in one set . checkpoint mrc-3d-dir/icos5f-refined-to-30.00Angstrom-fsc.txt . FSC . checkpoint mrc-3d-dir/All-particles-refined-to-30.00Angstrom-avg-ctf.agr . xmgrace file . average CTF weighted by the number of particles mrc-3d-dir/All-particles-refined-to-30.00Angstrom-ortcen-dist.agr . orientation of the particles on the asymmetric triangle mrc-3d-dir/All-particles-refined-to-30.00Angstrom-yields.agr . histogram of the percentage of particles used from each defocus mrc-3d-dir/icos5f-sym-thres.mrc File formats PRoot/jj0198/mrc-workdir/jj0198-refined-to-res-21.60Angstrom.dat /home/juan/PRoot/jj0198/mrc-workdir/jj0198-19.tnf 0, 72.2500, -0.8800, -35.5400, 361.1343, 360.0784, 0.00 PR(3[1/504A]-70[1/22A])=41.7,63.3, change: (0.00, 0.00, 0.00, -0.25, 0.00) Every pair of lines refers to a particle from the micrograph 'jj0198'. jj0198-19.tnf: the file containing the FFT of this particle. 0: not used 72.2500, -0.8800, -35.5400: Euler angles. 361.1343, 360.0784: y x 0.00: not used 3[1/504A]-70[1/22A]: low-high resolution in Pixel[spatial freq] 41.7: Cross-common lines phase residual 63.5: Self-common lines phase residual 0.00, 0.00, 0.00: change in Euler angles from the last iteration -0.25, 0.00: change in from the last iteration