Transcription data analysis code from the arXiv:1701.06079 manuscript.
To launch the code, open the main.py file. The lines of code are accompanied by comments for an easier understanding of the function. No installation is required to launch the code.
Every function located in a different file normally has a description at the top of the file. Most constants re-used throughout the analysis can be found and modified in the constants.py.
When running the main.py file from the terminal, comment out the first try... except... block, which provides code reloading for interactive environments like Jupyter.
To use the code, you need to supply the input data in a form of a table with pre-defined columns. The .csv file must be a comma-separated UTF-8 encoded file. An example of the input file called example_data.csv can be found in the example folder. The following columns are required to execute the code.
| Column | Description |
|---|---|
| time | Time in seconds |
| frame | An integer corresponding to the frame number of the recorded video. Used to detect the time step |
| intensity | Fluorescence intensity of the transcription region in the nucleus, in a.u. |
| ap | Location of the region on the anterior-posterior axis of the embryo, where ap=0 corresponds to the embryo's head |
| dataset | Dataset name. Can be any string not containing slashes or backslashes |
| dataset_id | A unique sequential identifier with a one-to-one correspondence to the data set name. The counting starts from 0 |
| construct | The name of the gene construct in the data set. Each data set must contain data from a single construct. In the data sets of the paper cited above, the construct takes one of 3 values: bac, no_sh or no_pr |
| construct_id | A unique sequential integer identifier for the constructs, starting at 0 |
| gene | Gene name. Each data set must contain data from a single gene. In the data sets of the paper cited above, the construct takes one of 3 values: hb (hunchback), sn (snail) or kn (knirps) |
| gene_id | Same as construct_id, but for genes |
| trace_id | A sequential integer identifier for fluorescent traces (nuclei), starting at 0. Must be unique within each data set. Each data set may contain multiple traces |
Python 3
The nuclear cycles are detected in each data set by first selecting a background threshold, which separates the continuous trace into nuclear cycles above the threshold.
The default intensity threshold is defined by default_intensity_threshold in constants.py.
If a data set needs individual adjustments to the threshold, they must be put into the intensity_thresholds dictionary in constants.py.
The ncs are then numbered. By default, it is assumed that the last observed nc is nc14.
In some data sets, not all of the ncs have been recorded, or artefact ncs may be created by thresholding.
To correct for this problem, one may manually specify the number of the last nc in the data set by adding it into the last_ncs dictionary in identify_ncs function in calculate.py.
The initial slopes are fitted to all points (non-averaged) of a nuclear cycles, in the first several minutes from the beginning of the nuclear cycle.
The exact fit duration is defined by the slope_length_mins constant.
The procedure is discussed in the accompanying article.
The nc must contain at least 3 frames to be fitted.
If the slope fit is negative, a nan value is stored instead.