barplots 1.1.30

Python package to easily make barplots from multi-indexed dataframes.

Python package to easily make barplots from multi-indexed dataframes.

How do I install this package?

As usual, just download it using pip:

pip install barplots

Documentation

Most methods, in particular those exposed to user usage, are provided with doc strings. Consider reading these docstrings for learning the most recent updates to the library.

Examples of the DataFrame structure

The dataframe to be provided to the barplots library may looks like the following:

miss_rate	fall_out	mcc	evaluation_type	unbalance	graph_name	normalization_name
0.0332031	0.705078	0.353357	train	10	AlligatorSinensis	Traditional
0.240234	0.478516	0.289591	train	1	CanisLupus	Right Laplacian
0.0253906	0.931641	0.101643	train	100	AlligatorSinensis	Right Laplacian
0.121094	0.699219	0.220219	train	10	HomoSapiens	Traditional
0.0136719	0.292969	0.722095	test	1	CanisLupus	Right Laplacian
0.0605469	0.90625	0.0622185	test	10	AmanitaMuscariaKoideBx008	Traditional
0.0078125	0.4375	0.614287	train	100	AmanitaMuscariaKoideBx008	Traditional
0.171875	0.869141	-0.0572194	train	100	AlligatorSinensis	Traditional
0.0859375	0.810547	0.150206	train	10	MusMusculus	Right Laplacian
0.0273438	0.646484	0.415357	test	10	MusMusculus	Right Laplacian

Specifically, in this example, we may create bar plots for the features Miss rate, fallout and Matthew Correlation Coefficient by grouping on the evaluation_type, unbalance, graph_name and normalization_name columns.

An example CSV file can be seen here.

Usage examples

Here follows a set of examples of common usages. Basically, every graph show either the same data or a mean based on the provided group by indices. Choose whetever representation is best for visualizing your data, as hardly one is better than another for every possible dataset.

N.B. The data used in the following examples are randomly generated so to be useful for test porposes. DO NOT consider these values as valid results for experiments using the same labels (cell lines etc…) which are only used to show possible usages.

For every example, the considered dataframe df is loaded as follows:

import pandas as pd

df = pd.read_csv("tests/test_case.csv")

Also, for every example, the custom_defaults used to sanitize the labels specific to the used dataset is:

custom_defaults = {
    "P": "promoters",
    "E": "enhancers",
    "A": "active ",
    "I": "inactive ",
    "+": " and ",
    "": "anything",
    "Validation":"val"
}

Horizontal Example A

In the following example we will plot the bars horizontally, rotating the group labels by 90 degrees and displaying the bar labels as a shared legend.

from barplots import barplots

barplots(
    df,
    groupby=["task","model"],
    orientation="horizontal",
    show_legend=True,
    minor_rotation=90,
    custom_defaults=custom_defaults
)

Result can be seen here.

Horizontal Example B

In the following example we will plot the top index as multiple subplots with horizontal bars, rotating the group labels by 90 degrees and displaying the bar labels as a shared legend.

from barplots import barplots

barplots(
    df,
    groupby=["cell_line", "task","model"],
    orientation="horizontal",
    show_legend=True,
    subplots=True,
    minor_rotation=90,
    custom_defaults=custom_defaults
)

Horizontal Example C

In the following example we will plot horizontal bars, rotating the top group labels by 90 degrees and displaying the bar labels as minor ticks.

from barplots import barplots

barplots(
    df,
    groupby=["task","model"],
    orientation="horizontal",
    show_legend=False,
    major_rotation=90,
    custom_defaults=custom_defaults
)

Result can be seen here

Horizontal Example D

In the following example we will plot the top index as multiple subplots with horizontal bars, rotating the group labels by 90 degrees and displaying the bar labels as minor ticks.

from barplots import barplots

barplots(
    df,
    groupby=["cell_line", "task","model"],
    orientation="horizontal",
    show_legend=False,
    major_rotation=90,
    subplots=True,
    custom_defaults=custom_defaults
)

Vertical Example A

In the following example we will plot the bars vertically and displaying the bar labels as a shared legend.

from barplots import barplots

barplots(
    df,
    groupby=["task","model"],
    orientation="vertical",
    show_legend=True,
    custom_defaults=custom_defaults
)

Result can be seen here

Vertical Example B

In the following example we will plot the top index as multiple subplots with vertical bars and displaying the bar labels as a shared legend.

from barplots import barplots

barplots(
    df,
    groupby=["cell_line", "task","model"],
    orientation="vertical",
    show_legend=True,
    subplots=True,
    custom_defaults=custom_defaults
)

Vertical Example C

In the following example we will plot vertical bars, rotating the minor group labels by 90 degrees and displaying the bar labels as minor ticks.

from barplots import barplots

barplots(
    df,
    groupby=["task","model"],
    orientation="vertical",
    show_legend=False,
    minor_rotation=90,
    custom_defaults=custom_defaults
)

Result can be seen here

Vertical Example D

In the following example we will plot the top index as multiple subplots with vertical bars, rotating the minor group labels by 90 degrees and displaying the bar labels as minor ticks.

from barplots import barplots

barplots(
    df,
    groupby=["cell_line", "task","model"],
    orientation="vertical",
    show_legend=False,
    minor_rotation=90,
    subplots=True,
    custom_defaults=custom_defaults
)

Future features

Currently it is not possible to automatically detect labels overlap and rotate them. I will work on this feature when I get the time, currently you will need to use the parameters minor_rotation and major_rotation.