Julia is a high-level language targeted at scientific data analysis and statistics. It shares similarities with both Matlab and Python in syntax, but can compile very fast (C-like) code when it can figure out that variables always contain primitive types (doubles, etc.).
Thus, a long-term goal is for the Julia implementation to be fast. It is presently a work in progress, however, and speed presently takes third priority behind existence and correctness of capabilities.
This code has been tested with Julia 0.4.2. It is no longer being tested with Julia 0.3.
To run this project you need to have JSON.jl installed.
Just type Pkg.add("JSON") at the Julia prompt.
The minimal WCON reader is in Minimal.jl. If you want to build a lightweight WCON reader/writer
for your own data, this may be a good place to start. To use it, just include("Minimal.jl") and then
enter
TrackerCommonsMinimal.read_wcon("my/path/to/my_data.wcon")to get a WormDataSet structure which roughly reflects the minimal WCON data features.
If you have a WormDataSet, you can write it out with
TrackerCommonsMinimal.write_wcon("my/path/to/new_file.wcon")TODO: write examples.
Worm data, including a single ID, plus a time series and x- and y-coordinate values, is
represented in the CommonWorm type.
TODO: finish section
Julia contains a Data Frame type, motivated by a similar type in R, that consists of tabular data with named column headers. Data frames are commonly used in data analysis and statistics, so it would seem that they would be a natural fit for worm tracking data.
However, tracking data is not necessarily a convenient rectangular shape: worms may be tracked for different periods of time, and within one worm, some data may be static (which side is ventral), some may be a scalar for each time point (centroid), and other data may be vector valued (x coordinates of spine). Forcing the data into tabular form could thus be inefficient and require the user to deal with frequent missing data.
Thus, the Tracker Commons implementation imports the data in custom structures. Users are encouraged to convert this to Data Frames in those cases where it suits their analysis.
TODO: finish section