lankycyril/pyvenn

Python module for plotting Venn diagrams of 2..6 sets

pyvenn: Venn diagrams for 2, 3, 4, 5, 6 sets

Please refer to the Jupyter notebook for demos and a brief explanation of the interface; a more complete documentation is in the works as the project keeps evolving:
https://github.com/LankyCyril/pyvenn/blob/master/pyvenn-demo.ipynb

This library is an evolution of tctianchi's pyvenn package (see fork URL).
Their liberal license (Unlicense) allowed me to fork the repository, change the license to GPLv3, modify the package's interface and, hopefully, significantly contribute to and improve the library, and make it installable from PyPI.

The main methods in this version are different from the ones in tctianchi's implementation, but the original methods are still provided for backwards compatibility, and I would like to emphasize the importance of tctianchi's work that allowed for this library to exist (among other things, figuring out the coordinates best fit for plotting the diagrams' shapes and petals' labels).

This iteration of the library implements two main functions:

• venn(dataset_dict, **kwargs) which plots true Venn diagrams for any number of sets between 2 and 5 using ellipses, and for 6 sets using triangles
• pseudovenn(dataset_dict, **kwargs) which plots a Venn-like intersection of six circles (not all intersections are present in such a plot, but many are).

pyvenn's Issues

Error message

Trying to run this in PyCharm. Says "NameError: name 'venn' is not defined." All packages seem to be installed. Also, I am new at python so I am unsure of some things, but how can I run my own data in this code?

Examples give 'dict_keys' object is not subscriptable

Try the examples in the README. They all have this error with version 0.1.3

Changing Graph SIze

When resizing with figsize, the image itself does not get larger than a certain amount, only the whitespace around it.

Is it possible to increase the image size without compromising the resolution? Generating labels for the petals gets squishy when more information is needed.

How to use a non-interactive backend?

How to use a non-interactive backend?

ax.get_legend_handles_labels() returns emtpy lists

I'd like to change the number of columns in the legend so that there is only one row (ncol = number of groups).

To do so I thought I could take the ax returned from venn and use the method .get_legend_handles_labels() which I could pass to ax.legend to also specify ncol. However, ax.get_legend_handles_labels() returns two empty lists, which when passed to ax.legend cause the legend to disappear.

musicians = {
    "Members of The Beatles": {"Paul McCartney", "John Lennon", "George Harrison", "Ringo Starr"},
    "Guitarists": {"John Lennon", "George Harrison", "Jimi Hendrix", "Eric Clapton", "Carlos Santana"},
    "Played at Woodstock": {"Jimi Hendrix", "Carlos Santana", "Keith Moon"}
}
ax = venn(musicians)
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles, labels, ncol=3)

The potential work-around is to just pass the keys of the dictionary musicians to ax.legend and also specify ncol (this produces a figure as expected), but still not sure why ax.get_legend_handles_labels() doesn't work.

sizing

Loving how much more user friendly your package is over the alternatives!

Is there a way to make the size of the circles (at least for 2 and 3 comparisons) based on the size of the sets?

Is this pip-installable?

Can I install this using pip?

pyvenn with matplolib subplots

Hello.
Recently I was trying to plot a Venn diagram in a matplotlib subplot but it always goes out of the plot. Is there a way to insert it back?

non-interactive/batch example to image file?

Can you create or describe an example that produces an image file? I'd like to leverage your work to quickly produce diagrams, but I'm not a matplot user, and I don't see where any file is written out.

For context, I tried:

pip3 install --user venn 
grep -v '^%' your-minimal-musician-example | tee musician.py
  from venn import venn

 musicians = {
     "Members of The Beatles": {"Paul McCartney", "John Lennon", "George Harrison", "Ringo Starr"},
     "Guitarists": {"John Lennon", "George Harrison", "Jimi Hendrix", "Eric Clapton", "Carlos Santana"},
     "Played at Woodstock": {"Jimi Hendrix", "Carlos Santana", "Keith Moon"}
  }
  venn(musicians)

python3 musician.py

The file runs once the %matplotlib inline is removed, but the result doesn't export the generated object as an image. Perhaps it's trying to manipulate a running matplot process?

Venn() does not generate image

I follow the minimal example in the description page, but when I run the module (IDLE), though it throws no error, it does not generate any image.

'module' object is not callable

I try to import venn but it doesn't work

When I try to import venn:
from venn import venn
I have the next error:

import venn
ModuleNotFoundError: No module named 'venn'

Export counted values in petals

Hi,

pyvenn is very useful but for my use it lacks an export function of information on the groups in each petal.
The code below is a suggestion for this type of function. It exports the results as a table (dataframe).
You can improve it, if you don't want to use pandas.

import pandas as pd
from venn import *

def generate_logics(n_sets):
    """Generate intersection identifiers in binary (0010 etc)"""
    for i in range(1, 2**n_sets):
        yield bin(i)[2:].zfill(n_sets)
        
def generate_group_labels(datasets, fmt="{size}"):
    """Generate petal descriptions for venn diagram based on set sizes"""
    datasets = list(datasets)
    n_sets = len(datasets)
    dataset_union = set.union(*datasets)
    universe_size = len(dataset_union)
    petal_labels = {}
    petal_set2 = {}
    for logic in generate_logics(n_sets):
        included_sets = [
            datasets[i] for i in range(n_sets) if logic[i] == "1"
        ]
        excluded_sets = [
            datasets[i] for i in range(n_sets) if logic[i] == "0"
        ]
        petal_set = (
            (dataset_union & set.intersection(*included_sets)) -
            set.union(set(), *excluded_sets)
        )
        petal_labels[logic] = fmt.format(
            logic=logic, size=len(petal_set),
            percentage=(100*len(petal_set)/max(universe_size, 1))
        )
        petal_set2[logic] = fmt.format(
            logic=logic, size = (dataset_union & set.intersection(*included_sets)) -
            set.union(set(), *excluded_sets)
        )
        labelset = pd.DataFrame(petal_labels.items())
        grpset = pd.DataFrame(petal_set2.items())
        dfset = labelset.merge(grpset, on = 0)
        dfset.columns= ["Comparison", "Count", "Groups"]
    
    return dfset

Example:

groupset = {
    "Cells 1": {'RBM15', 'FAM76B', 'TNIK', 'TUSC3', 'ZFYVE27'},
    "Cells 2": {'RBM15', 'TMEM8A', 'ARRDC2', 'TUSC3', 'ZFYVE27'},
    "Cells 3": {'ARRDC2', 'UNC50', 'TNIK', 'TUSC3', 'TMEM8A'},
    "Cells 4": {'RBM15', 'UNC50', 'TNIK', 'TUSC3'}
}

dfset = generate_group_labels(groupset.values())
dfset

Output:

Comparison	Count	Groups
0001	0	set()
0010	0	set()
0011	1	{'UNC50'}
0100	0	set()
0101	0	set()
0110	2	{'TMEM8A', 'ARRDC2'}
0111	0	set()
1000	1	{'FAM76B'}
1001	0	set()
1010	0	set()
1011	1	{'TNIK'}
1100	1	{'ZFYVE27'}
1101	1	{'RBM15'}
1110	0	set()
1111	1	{'TUSC3'}

Best regards,

JR

Documentation on Changing Colors

I'm trying to figure out a way to specify specific color mappings. I've put a list of rgb colors into a 2D list but it's failing to take the 2D list and parse it.

Is there another way to specify colors. I have to use specific colors for work.

Allow `legend_loc=None`

As title says please.

How to show names to the sets on the plot?

Dear all,

How to show names to the sets on the plot?
If there are 4 sets say A, B, C and D, how to show them on the sets in the resulting figure?

Thanks,
Burcak