Leginon Developer Guide¶

Overview of the code repository¶

Over the years, Leginon has grown and divided. With the addition of the Appion processing pipeline, several components that were once subsystems of Leginon were broken out to be used as a common tool set for both Leginon and Appion. The top level master package that now contains the entire software suite developed at AMI is called myami. Inside of myami you will find leginon, appion and all of these common packages. Here is a brief description of the contents of the AMI software repository:

• myami - the complete suite of software
  • appion - the processing pipeline
  • comarray - Python extension to retrieve COM SafeArrays into NumPy arrays (deprecated, see Installing comtypes instead)
  • dbschema - If you are running your myami system out of a sandbox and frequently pulling changes from the repository, then your database schema is generally kept up to date automaticall. But most production installations update only between major revisions. The dbschema directory summarizes all of the database schema updates between revisions, which can be applied in sequence when doing a major jump in revisions.
  • imageviewer - this is a newer image viewer widget for wx meant to replace the old ImageViewer widget in the Leginon GUI. So far only the tomography node uses it.
  • install - adds automation to the installation process
  • leginon - microscope control and automation
  • modules - python extensions that must be compiled
    • libcv - key point detection and matching
    • numextension - misc image functions. More info here: numextension
    • radermacher - functions for tilt picker, see http://emg.nysbc.org/projects/tiltpicker/wiki/TiltPicker_Installation
  • myamiweb - all of the php web server stuff for the the web viewer, appion pipeline, etc.
  • programs -
  • pyami - miscellaneous pure python modules that are usefull for both leginon and appion
  • pyscope - Python interface to several TEMs and cameras
  • redux - a Python replacement for php_mrc, but also generally useful on its own for image conversions, FFTs, resizing, etc.
  • sinedon - object relational mapping module, so you can use python objects to populate your MySQL database

Brief description of the design of Leginon¶

Leginon is a flexible framework for automating your TEM data collection. The flexibility comes from the fact that key tasks are coded into nodes, which can then be dynamically assembled into an event driven application. Each node performs a task, and generates events to trigger other nodes to perform their tasks. In addition to passing events between each other, nodes may also share the data they produce. For example, an Acquisition node may acquire an image from the instrument, then generate an event to notify another node to process that image. Another area of flexibility is in how applications may distribute the nodes among several hosts on the network. Processing could be optimized depending on which nodes are on the instrument host versus which ones are on a more powerful host elsewhere.

Adding new features and functionality to Leginon¶

The two most common ways you will be adding a new idea to Leginon:

• A small addition to an existing node. For example, a new parameter to auto focusing would require a new entry box in the GUI and a modification to some underlying function call.
• A completely new feature that is different enough to belong in its own node. For example, a new target finding technique. See Creating a new Leginon node

Simulation: when you do not have access to a real TEM and/or camera.¶

• You still need the overall infrastructure in place: myami installed, database created, etc., but at least this can still be an option for a stand-alone laptop that is not even connected to the network.
• configure your instruments.cfg file with SimTEM and SimCCDCamera instruments. These are the defaults in instruments.cfg.template.
• Start Leginon and do not connect to any external clients during the session setup wizard
• Go through calibrations as if they are real microscope and camera. You can however, fake the values.
• Most functionality can be tested, however your images will be simulated and the more advanced target finders may not work so well.
• Premade Realistic-looking calibration and images for simulation

Useful diagonosis scripts¶

• testing pyscope instrument functions without Leginon.
• setup simulator with VMware