Tutorial: Ab(40) using Frealix

Written by Erin Fan, Last modified: 02/28/2013

Edited by X

 

 

 

 

 

 

 

 

  1. Picking the Filaments
    In a directory named "boxes", pick and create boxer files for all of the good filaments in the films.

     

    • Create a directory named "boxes".
    • Open each film with EMAN Boxer using the command boxer file.mrc.
    • Check and correct the contrast of the filaments against the film.
    • Using the right mouse button, move the film to search for a good filament. It is best to scan- through the film in a systematic manner such that no filaments will be missed.
    • With "Select" selected on the tools window, begin by picking an endpoint. This endpoint (as shown by the green arrow) should be located along the filament near its starting end, and should not be a crossover.
    • Following the filament, select the next crossover points in order.
    • After the last crossover has been selected, pick an endpoint near the end of the filament.
    • When a full filament has been selected, click on Boxes -> Save Box DB in the tool box (or Ctrl+S) to save your box file in the "boxes" directory.
    • After the file has been saved, click on Boxes -> Clear Boxes in the tool box.
    • Repeat the previous steps until all good filaments in the films have been picked.
  2. Contrast Transfer Function (CTF) Correction

    The following 2 steps, Preliminary 3D Reconstruction with a single filament as well as Refinement of the model with ten filaments, may be executed automatically by running the step_by_step.txt script.

    • A new directory named "ctf" is created.
    • By running the ctffind script, each micrograph now has a ctffind log file.
  3. Creating a Preliminary 3D Reconstruction with One Filament
    In a different directory (named "fil11"), a random filament (#11) is selected and, without using a model, one round of Frealix is run to create a 3D reconstruction.

     

    • The micrograph film (.mrc) of the selected filament and its corresponding boxer (.box) file are copied to the directory, along with the wrapper script (flx_wrap.py), a new flx file, and a link to the Frealix executable.
    • The flx file should have DS_RANDOM_START = T and NUM_MINIM_RUNS = 10. (Note: Make sure correct paths were set for INPUT_BOX_FILES, FILM_FILES, and CTFFIND_FILES).
    • Making use of the wrapper script (flx_wrap.py), the command nohup ./flx_wrap.rb -r 1 -s 20 -f 20 -t 20 -n 1 fil_1auto.flx | tee flx_1auto.log is used to run a round of Frealix with 10 random starts on the cluster with one node.
    • When finished, you should have frealix_round_000, frealix_round_001 (with 3drec.mrc), and a log file.
  4. Refining the Model with Ten Filaments
    In another directory ("tenfils"), nine more filaments are chosen randomly (filament 1, 8, 10, 14, 17, 168, 188, 191, 212). Using the model created from one random filament (#11), as shown above, four rounds of Frealix are run with a total of ten filaments (the selected filament #11 and the other nine randomly-selected filaments).

     

    • Copy the 3drec.mrc model from the latest round (frealix_round_001) of the fil11 directory to here.
    • The micrograph films (.mrc) of the selected filaments and their corresponding boxer (.box) files are copied to the directory, along with the wrapper script (flx_wrap.py), the old flx file, and a link to the Frealix executable.
    • The old flx file is edited and saved as 10fil.flx. Changed parameters include updating the locations of the INPUT_BOX_FILES and FILM_FILES, adjusting the FILM_SELECTION to include the corresponding films of the selected filaments (film #1, 2, 22, 25, 26, 30), and setting a new parameter PSF_RADIUS to 400.
    • The command nohup ./flx_wrap.py -n 1 -s 20 -f 20 -m 3drec.mrc --num_nodes=10 10fil.flx | & tee flx_yymmdd.log is used to run one round of Frealix on the cluster with 10 nodes.
    • In the flx file, switch the WPS_ARE_CROSSOVERS flag from T to F. Again using the wrapper script, run 3 additional rounds of refinement with nohup ./flx_wrap.py -n 3 -s 20 -f 20 --num_nodes=10 10fil.flx | & tee flx_yymmdd.log.
  5. Using the Model to Refine the Full Bin8 Data
    The newly-obtained 3drec.mrc model from frealix_round_004 is then copied to a full bin8 data directory. Using the model, the first round of refinement is done.

     

    • The wrapper script (flx_wrap.py), a flx file, and a link to Frealix are copied to a new directory, named full_bin8. (Note: Make sure correct paths were set for INPUT_BOX_FILES, FILM_FILES, and CTFFIND_FILES).
    • The command nohup ./flx_wrap.py -n 1 -s 20 -f 20 -m 3drec.mrc --num_nodes=350 bin8.flx | & tee flx_yymmdd.log is used.
    • The resulting directory should have a frealix_round_000 and frealix_round_001.
  6. Continuing to Refine the Full Bin8 Data
    In the parameters file, switch the WPS_ARE_CROSSOVERS flag from T to F. Again using the wrapper script, run four additional rounds of refinement.

     

    • The command nohup ./flx_wrap.py -n 4 -s 20 -f 20 --num_nodes=350 bin8.flx | & tee flx_yymmdd.log is used.
  7. Unbinning the Bin8 Model
    Using Frealix's BIN mode, the bin8 3D reconstruction from the last round was unbinned 2x to bin4.

     

    • ./frealix2
      Please give FREALIX_MODE: BIN
      Please give INPUT_FILENAME: frealix_round_005/3drec.mrc
      Please give OUTPUT_FILENAME: 3drec_bin4.mrc
      Please give BINNING_FACTOR: 2
      Please give UNBIN: T
      Please give CORRECT_ORIGIN: F
  8. Using the Model to Refine Full Bin4 Data
    The bin4 model is moved to a new directory named full_bin4 with the latest waypoints.ali file. Here, ten rounds are run in order to achieve a resolution closer to 10A.

     

    • The waypoints.ali file is moved into a frealix_round_000 directory created in full_bin4.
    • The command nohup ./flx_wrap.py -n 1 -s 20 -f 20 -m 3drec_bin4.mrc --num_nodes=350 bin4.flx | & tee flx_yymmdd.log is used to run one round of refinement.
    • Edit the flx file such that POLARITY_CHECK is switched to F.
    • Six more rounds of refinement from 18Å to 12Å are run using the command nohup ./flx_wrap.py -n 6 -s 18 -f 12 --num_nodes=350 bin4.flx | & tee flx_yymmdd.log.
  9. Unbinning the Bin4 Model
    Using Frealix's BIN mode, the bin4 3D reconstruction from the last round was unbinned 2x to bin2.

     

    • ./frealix2
      Please give FREALIX_MODE: BIN
      Please give INPUT_FILENAME: frealix_round_007/3drec.mrc
      Please give OUTPUT_FILENAME: 3drec_bin2.mrc
      Please give BINNING_FACTOR: 2
      Please give UNBIN: T
      Please give CORRECT_ORIGIN: F
  10. Using the Model to Refine Full Bin2 Data
    The bin2 model is moved to a new directory named "full_bin2" with the latest waypoints.ali file. Here, five rounds are run in order to achieve a higher resolution.

     

    • The waypoints.ali file is moved into a frealix_round_000 directory created in full_bin2.
    • Note that in bin2.flx, REFINE_WPS_SIMULTANEOUSLY = T, and the SEARCH_HALF_RANGE options have been removed.
    • Five rounds of refinement from 12Å to 8Å are run using the command nohup ./flx_wrap.py -n 5 -s 12 -f 8 --num_nodes=350 bin2.flx | & tee flx_yymmdd.log.

Note: yymmdd should be replaced with today's date.