Automation of jeol1230 with iRobot, Leginon and Animated-TEM

Contents

Startup

User manual to use Leginon combined with the Matlab nodes associated to the Animated-TEM Toolbox algorithms at NYSBC: NYSBC-AutoRuns_Protocols_4.doc.

Tracking Developments

Installation

Matlab applications under Leginon

• v0.2 works with target selections at low and medium magnifications using Matlab. The filtering nodes are however useless. The magnifications are the same as in the "Robot MSI-Screen 1st Pass 1.6" application.
• v0.3:
  • Nodes have new specific names.
  • Filtering nodes removed.
  • High Mag processing added (with a dummy "Very High Mag" node after it to avoid errors).
• v0.4b:
  • Nodes have new specific names.
  • Focusing added before High magnification

Leginon program modifications

!! THE FILES MODIFIED HAVE ONLY BEEN UPDATED ON CEM12 SO FAR, AND NOT ON JEOL1230 (as of Feb, 24th) !!

for communication with Matlab

• original file: matlabtargetfinder.py
  • 1st modif (matlabtargetfinder.py.MODIF_OK_1): convert images to float before sending them to Matlab (initially saved as "int", the mlabwrap interface does not recognize this type and send zeros instead of the true matrix)
  • 2nd modif (matlabtargetfinder.py.MODIF_OK_2): same as version 1, but more complete (works wether the prog is called off-line on a test image or on-line).

to send a new input to Matlab

• previous vcersion: "matlabtargetfinder.py.MODIF_OK_2" file
  • 3rd modif (matlabtargetfinder.py.MODIF_OK_F2): send image ID to Matlab as well + code to transform from int to float better (from Jim and Anchi's advices)

to modify the GUI of Leginon (Set Parameters) so that a Matlab GUI can be launched to parametrize the algorithms (see http://ami.scripps.edu/redmine/boards/6/topics/884?r=890#message-890)

• original file: "leginondata.py.ORIG"
  • 1st modif: leginondata.py.MODIF_OK_1; new input declared (parametergui path)
• and original file: "/gui/wx/MatlabTargetFinder.py.ORIG"
  • 1st modif: "/gui/wx/MatlabTargetFinder.py.MODIF_1_OK": add components to browse the Matlab GUI, and a button to launch it.

autofocusing -- to reduce false correlation peak detection during the autofocusing steps in cross-correlation mode (phase-correlation not corrected)

• original file: "calibrationclient.py.ORIG"
  • modif: "calibrationclient.py.MODIF_1" The file has been modified to activate a Gaussian smoothing before searching for the peak (in the def measureZ function, shiftinfo = self.measureScopeChange(im1, state2, correlation_type=correlation_type, lp=1) changed to shiftinfo = self.measureScopeChange(im1, state2, correlation_type=correlation_type, lp=1)), and do a median filter for "isolated" pixels with high gray level .

example of command to update a python file of Leginon:

sudo cp /home/ncoudray/Desktop/Matlab_leginon_protocol/matlabtargetfinder.py.MODIF_OK_F2 /usr/lib/python2.4/site-packages/leginon/matlabtargetfinder.py

autofocusing -- Beam-tilt method

• jeol1230.py, line 333 corrected from "_return Ture_" to "_return True_"
• Beam tilt for jeol1230 has not been implemented... For the moment, no conversion has been done from radian jeol's decimal value. To see a beam tilt, tilt value should be above 1000. Calibration has been done with 5000. During the run, for the moment, 2000 can be used.

autofocusing -- New method based on Contrast Function * Measure the relative contrast of images taken around current focus and look for defocus at which the contrast function is maximized: cf modification in the file Focuser.py

Animated-TEM-4L

Animated-TEM-4L (ANalysis of IMages for Automatic Targeting and Extraction of Data in TEM - For Leginon) is adapted from Animated-TEM. It does not work as "master" which decides when, where and how images should be acquired. Therefore, "Anim_Scenario.m" previously used to call the different algorithms and program is not used anymore. Instead, it launched by Leginon when needed and outputs its target selection. Algorithms are quite similar (a few adaptations concerning the parameters and the methods have been done). GUI are not longer usable "as they are".

Specific ".m" file have been created for each magnification: these are the files called by Leginon and they link to the required algorithm, embedding it with the required inputs and managing the outputs (saving and sending the appropriate to Leginon):

• at low magnification algorithm, the intermediate file is called, for the moment (Feb 04), "Test_LowMagTargeting_v0_1.m".
• at medium magnification algorithm, the intermediate file is called, for the moment (Feb 04), "MedMagTargeting_v0_1.m".

Saving the extracted data:
Also, the way the data were saved had to be changed:

• initially, one folder was dedicated to images, logfiles and data. In Animated-TEM-4L, we use a mysql database. "SaveLowMagData.m" and "SaveMedMagData.m" have been completed to save data extracted by the algorithm into a new mysql database (One new input "TVersion" tells which way the data must be saved. It has been rewritten so that these two files can be used by both Animated-TEM. "Anim_Scenario.m" has been corrected to include this new "TVersion" input).
• The database is called "AnimatedTem" (see below for more information).

Management of the parameters:

• Specific parameters are still located in "AllInitialParameters.m". They have been adapted to suit images acquired so far with the jeol1230.
• Other user-changeable parameters were previously managed through "AnIm_Param.m" which created a "Acq_Param.txt" file specific for each run according to the settings chosen by the user through the Launcher GUI. Instead, a new strategy has been chosen here: "AnIm_4L_Param.m" file replaces "AnIm_Param.m". Some parameters have been removed (mainly those specific to the command of the microscope), others remain quite similar (those related to the algorithms), other have been added (those related to the databases). Furthermore, instead of saving the current parameters in a text file, a mysql datase called "AnimatedTem" has been created (on cem100, see next point).

GUI:

• "Anim4L_Calibration.m": replace the launcher to changed the default values for the parameters used by the algorithm. When saved, data are placed at the last row of the RunParameters table of the AnimatedTem database.
• "Grid_N_Cassette_Overview.m": displays, for a given session, the results saved in the AnimatedTem database.

Some algorithm modifications:

• Low Mag
  • Jan-Feb 2011: slight auto-threshold adaptation (SortAllSquaresGrayLevels.m file)
  • March 3rd 2011: Pause the run if superposed grids are detected (LowMagnificationSpeed_AnIm.m file)
• Med Mag
  • Jan-Feb 2011: Adaptation of the parameters for the automated process (AllInitialParameters.m file_: scales used, edge filter) + correction of "parameters checking" in MultiScale_vMP3.m [cf ThreshAlgo variable]
  • Feb 2011: do not restrain research to monolayer. Bi-Layer also included under certain conditions
  • Feb 2011: in MultiScale_vMP3.m , the way to measure
    • the background percentage slightly changed (a second background threshold is included)
    • the non-stacked percentage slightly changed (mono-layer & partly bi-layer)
    • the average rank of each regions slightly changed (option "RankProportional" to include a proportionality of adjust the rank according to the importance of the parameter's value)
  • Mar 2011: in MultiScale_vMP3.m, the way to remove the false edges:
    • Threshold (OPTIONS.Thresh) moved back from 10 to 8 (less edges removed)
    • Do not remove edges iteratively if more than 2000 regions are identified (save processing time)

Matlab database

General information:

• located on cem100
• name: AnimatedTem
• user name: usr_object
• password:
• driver: com.mysql.jdbc.Driver (must have been installed on the cem12)
• url: jdbc:mysql://192.168.5.233/

Tables and columns (as of 2011/02/04):

• RunParameters: store parameters used by the algorithms. Parameters can be changed by the user (see above). By default, the algorithms always consider the last entry of the table as the one which should be used!!!! Name of the columns begin with LM when the parameter is related to the algorithm used at low mag, with MM for the algo used at medium mag, with HM for the algo used at high mag.

• • if other parameters than the ones above are required to set the algorithm:
    • then, the required columns should be added in the database.
    • "AnIm_4L_Param.m" file should be completed so that, when it reads the RunParameters table of the database, the added columns are recognized (once checked, parameters are transmitted to the different algorithm through a structure called "Parameters". To avoid adding additional inputs and outputs, just complete this structure (see code for examples)).
  • to add the algorithm itself:
    • Choose a name and complete the "AnIm_4L_Param.m" file so the name of your algo is recognized (complete the "AcceptedValues" cell with this name for the appropriated magnification).
    • Add the link to your file with the appropriate name chosen above, the appropriate inputs and outputs in the corresponding file (LowMagTargeting_, MedMagTargeting_, or HighMAgAnalysis_; see instructions around "if strcmp...").
  • to save new outputs in the database:
    • Complete one of the tables below and the corresponding m file ("SaveLowMagData.m", "SaveMedMagData" or "SaveHighMagData")

• GridMag: store the data outputs by the algorithm at low mag

• MembraneMag: store the data outputs by the algorithm at medium mag

• [Another one should be created if diffraction patterns are analyzed at high magnification]: StructureMag:

• Set ALLOWTOPICVIEW =

-- NicolasCoudray - 14 Apr 2011