Python package to easily make barplots from multi-indexed dataframes.
As usual, just download it using pip:
pip install barplots
Since some software handling coverages sometime get slightly different results, here's three of them:
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.
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"
}
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.
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
)
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
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
)
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
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
)
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
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
)
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
.