A Python package to allow fast, modular and reproducible generation of publication quality scientific figures and movies with only a few lines of code.
Figureflow was used to generate all 26 figures (10 main, 16 supplementary) and 22 movies for Schelski and Bradke 2022:
https://www.science.org/doi/10.1126/sciadv.abo2336
FigureFlow was also used to generate all figures and all movies from the corresponding preprint Schelski and Bradke 2021 (examples from preprint at the bottom of this page):
https://www.biorxiv.org/content/10.1101/2021.09.01.458567v1.full
Figureflow also includes a GUI for image modifications (add zoom, arrow, text etc.) to generate code that can be copied and pasted into the script to apply the modifications:
figureflow_GUI_sped_up_repeated.mp4
A lot of the code to plot data and add significance information (in statannot.py) is from statannot https://github.com/webermarcolivier/statannot created by Marc Weber.
For any questions or an introduction of how to use the package please contact me via E-Mail to max.schelski@googlemail.com. I am very happy to help people starting to use the package and I am very open to any feedback - including anything not working properly or easily.
After installation (see section "Installation" below), start with looking at some of the many available examples for movies or figures. Examples of figures are under data/example_figures/ as .png files, while example movies are under data/example_movies as .mp4 files (for details see section "Usage Examples" below).
For questions about specific functions, check the API section of the documentation under http://figureflow.readthedocs.io/en/latest/api.html - all parameters for all functions are documented there.
If something is unclear, not well explained, you are unsure whether something is implemented or something doesn't work, feel free to write me an E-Mail to max.schelski@googlemail.com - I am happy to help.
- create figure according to exact size necessary for desired journal in inches
- automatically perfectly aligns plots and images in figure panels
- Display illustrations as png file; or as powerpoint file to unify font size
- ...
- Provide images as single Hyperstack or single images
- Define which timepoints, channels etc. to display
- Position of images is automatically determined... but can be customized as well
- Contrast set in ImageJ is automatically extracted from Hyperstack
- Add annotations (star, asterisk, arrow, lines,text), also using GUI
- Define regions to be zoomed in, crop images, also using GUI
- Add timestamp, scale bar, channel names
- Annotate outside of image based on dimension (e.g. Annotate on the outside all timepoints)
- ...
- Provide .csv file
- Plot data automatically in the allocated space (Boxplot with overlaying data points)
- Calculate significance and annotate in image (Largely based on Statannot from Marc Weber)
- Show statistics (e.g. Mean, median, SEM, etc) for each data group (Largely based on Statannot from Marc Weber)
- Show sorted list of representative cells (closest to average)
- ...
- Movies can be generated as full-fledged figure objects
- Images can be supplied as ImageJ Hyperstack with the "frames" attribute being animated
- Lineplots of quantifications (.csv file) can be animated based on the x-attribute
- Add title pages to movies with introductory text
- Precisely define movie quality (resolution and bitrate)
- ...
Some documentation of how to generate movies can be found in the folder "scripts" in create_movies.py.
Many usage examples covering almost all of the features of figureflow can be found in the the folder "scripts" for generating figures (under "scripts/figures") or movies (under "scripts/movies"). The figures and movies are from Schelski and Bradke 2022.
To find out which examples might be interesting to you, start by looking
- in "data/example_figures" for generated figures (as .png file, e.g. data/example_figures/figure1.png) or
- in "data/example_movies" for generated movies (as .mp4 file, e.g. data/example_movies/MovieS1.mp4)
Check which examples contain features you would like to use. Then you can use the corresponding script for the example of interest (e.g. scripts/Figure1.py or scripts/MovieS1.py) to play around with the feature of interest and thereby understand the usage hands-on.
The data used for the figures and movies is in the folder "data" in a subfolder with the same name as the name of the script (e.g. the data for scripts/figures/Figure1.py is in data/Figure1).
The data for a figure can contain:
- A .csv file named __figure__ followed by the figure number (e.g. "__figure__S1.csv") file that describes the layout of the panels
- .csv files for analyzed data
- .tif files for microscopy images
- .png/.pptx files for illustrations.
Figureflow automatically assigns files to the respective panel by file name:
The name of files for a specific panel starts with "panel" followed by the uppercase panel letter and then an underscore (e.g. a file for panel A starts with "panelA_"). Alternatively, files for a panel can be stored in a folder named "panel" followed by the uppercase panel letter ("e.g. "panelA").
Installation via Mamba (using the conda-forge channel) is recommended, since Mamba is a much faster version of Anaconda.
- If you don't already have Mamba installed: Download and install Mamba from https://github.com/conda-forge/miniforge.
- If you don't already have git installed: Download and install git from https://git-scm.com/downloads
- Open a terminal, navigate to the folder where you want to put Figureflow and clone the Figureflow repository:
- Navigate into the folder of the repository (figureflow):
cd figureflow
- Create environment for Figureflow with Mamba:
mamba env create -f environment.yml
- Activate environment with mamba:
mamba activate figureflow
- Install FigureFlow locally using pip (also include the "." in the command):
pip install -e .
- Optional: If needed, install spyder to create, edit and run scripts:
mamba install spyder
- Optional: start spyder
spyder
- You can now import Figureflow and use it to build figures and generate movies as figure objects:
from figureflow.figure import Figure
- Open and execute example scripts from repository in the folders figureflow\scripts\figures and figureflow\scripts\movies.
For any questions feel free to contact me via E-Mail to max.schelski@googlemail.com.
Figureflow is constantly updated. For updating Figureflow on your machine, open a terminal, navigate into the folder where you put Figureflow and then type:
git pull
This will download all new files on GitHub (if there are some).
Note: However, since the package was developed in Windows, the exact package versions for the environment in Windows are also available (environment_windows.yml) for Anaconda (using the default Anaconda channels).
The following explains the general structure of scripts using figureflow.
However, it does not provide executable code. For executable code check the python files in the "scripts" folder of the repository. The upper section "Usage Examples" provides more details about the different provided examples.
In general you first create a figure object.
figure = Figure(...)
You should now create a .csv layout file (see section "Creating a layout for a figure" below). Then you can execute all functions on this figure object. You should first create a panel. For that, you first need to have a csv file (for showing data plots) or an image file (e.g. tiff; for showing images). The filename must start with panelA. If there is no file for the panel you want to create, an error will be raised.
figure.create_panel(letter="A",...)
This panel can either show images (if the panel files are image files) using:
figure.show_images(...)
or it can show data plots (if the panel file is a csv file) using:
figure.show_data(...)
The all functions and their parameters related to showing data plots are described under the "Dataplot panel" navigation point, while all functions and their parameters related to showing images are described under the "Image panel" navigation point.
The layout of a figure is created by defining the position and size of the panels. This can either be done by setting the parameters x, y, width and height separately for each figure panel or by creating a csv file the following way:
The csv needs to be in the main folder of the figure with the name figure followed by the "number" and ".csv" A grid with letters of the panels where if a panel should span several cells the corresponding letter should be written in several cells.
Widths and heights can be defined by two possible ways:
- In which the first row is the width of each column and the last column is the height of each row
- In which above every panel letters there is a number corresponding to the relative width of the panel in that row and the last column of the csv contains the height of each row
- In which widths are defined as in the second point but heights are defined for every panel by a number right of the panel letter
If you want to add text, arrows, zooms or cropping to an image in a panel, you can also use the GUI for one panel at the time. The GUI also allows you to adjust the position of elements you added already. With the GUI you don't need to check in ImageJ at which coordinates you want to add an arrow or text.
To use the GUI, right after defining the figure object (non executable code):
Figure(...)
you can write the following:
figure.edit_panel("C")
(where "C" is the panel you want to edit)
Adding arrows or text on images which will be cropped should only be done with setting the parameter "change_cropping" to False:
figure.edit_panel("C", change_cropping=False)
Otherwise, the image is shown without cropping applied and therefore the relative size of arrows or text compared to the image will be different than after cropping is applied.
After you are done adding and modifying the elements you want, you have to press the "Print code" button, which will print the code for the elements in your Python console. You can then copy and paste the code into the script for the figure/movie, making sure you paste each command at the right spot (before or after show_images and to the right panel).
Note, if some annotations were already present before adding annotations, the code for the annotations that were already present will also be printed. The reason is that those annotations might have been moved around and therefore changed.
Examples are not displayed to scale on this website but were all generated with 4.75 inch width.
Examples from Schelski and Bradke 2021 fully generated with the package without Adobe Illustrator etc:
