ezCharts

(Apologies to non-US English speakers).

ezCharts is a Python library for creating and rendering charts through eCharts or Bokeh. Plots can be constructed through an API similar to seaborn.

Additionally, ezCharts ships with a layout system built around dominate, providing a framework for creating static HTML reports via a declarative syntax.

Using the charting and layout functionality, a library of report components is provided in the domain of bioinformatics analysis and Nanopore sequencing.

Installation

ezCharts is easily installed in the standard Python tradition:

git clone --recursive https://github.com/epi2me-labs/ezcharts.git
cd ezcharts
pip install -r requirements.txt
python setup.py install

or via pip:

pip install ezcharts

Usage

The Plot() API in ezCharts mirrors the eCharts API in order that everything follows the eCharts documention. The API was in fact constructed from an API schema encoded in the source code of the documentation site. Users can therefore follow the eCharts documentation to construct charts with ezCharts. This differs from the pyecharts library which adds a layer of indirection.

eCharts plots

The library contains a Plot class for constructing plots using eCharts. Instances of this class have an attribute hierarchy that mirrors the eCharts Option API. Attributes can be set by providing a dictionary, runtime type checking ensures that child attributes match the Option API:

from ezcharts.plots import Plot
from ezcharts.components.reports.comp import ComponentReport

plt = Plot()
plt.xAxis = dict(name="Day", type="category")
plt.yAxis = dict(type="value")
plt.dataset = [dict(
    dimensions = ['Day', 'Rabbits'],
    source = [
        ['Monday', 150],
        ['Tuesday', 230],
        ['Wednesday', 224],
        ['Thursday', 218],
        ['Friday', 135],
        ['Saturday', 147],
        ['Sunday', 260]
    ]
)]
plt.series = [dict(type='line')]
ComponentReport.from_plot(plt, "tmp.html")

Up to the the final line, the code here mirrors exactly the javascript eCharts API. Note, many of the examples in the eCharts API set data items on the xAxis and series attributes. However the eCharts dataset documentation advises setting data within the dataset attribute; doing so provides an experience somewhat akin to ggplot2 in R or seaborn in Python. The primary use is to create additional datasets through data transforms:

plt = Plot()
plt.xAxis = dict(name="Day", type="category")
plt.yAxis = dict(type="value")
plt.dataset = [...]  # as above

plt.add_dataset({
    'id': 'filtered',
    'fromDatasetIndex': 0,
    'transform': [{
        'type': 'filter',
        'config': {'dimension': 'Rabbits', 'gt': 200}
    }]
})

plt.series = [dict(type='line', datasetIndex=1)]

The above example shows the use of a simple filter to plot only a subset of the data. More usually transforms can be used to plot multiple series based on a facet of the data.

The example also shows use of the convenience method .add_dataset(): this is provided to ensure the provided dictionary is type-checked against the eCharts API. The alternative would be to call .append({...}) on the plt.dataset attribute. However, this is at risk of error. Similarly, the .add_series() method exists to attach additional series to the chart.

Gotchas

It is not currently possible to set child attributes without first setting a parent, i.e. the following is not possible:

from ezcharts.plots import Plot

plt = Plot()
plt.xAxis.name = "My Variable"

This may change in a later release.

Rendering a chart may result in in JSON encoding errors. To resolve this, amendments are needed to excharts.plots._base to define how types can be encoded to JSON.

Bokeh plots

Since v0.6.0, Bokeh can be used as alternative plotting backend. To this end, a wrapper class BokehPlot exists. The Bokeh figure instance can be accessed under the _fig attribute (see example below).

from ezcharts.plots import BokehPlot
from ezcharts.components.reports.comp import ComponentReport

days = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
rabbits = [150, 230, 224, 218, 135, 147, 260]

plt = BokehPlot(
    x_range=days, title="some title", x_axis_label="Day", y_axis_label="Rabbits"
)
plt._fig.vbar(x=days, top=rabbits, width=0.9)

ComponentReport.from_plot(plt, "tmp.html")

Seaborn API

A higher level API is provided that mirrors the seaborn API to allow creation of common plot types without knowledge of eCharts. This currently has minimal functionality that will be added to over time. The idea is that eventually most plotting can be performed through this API without requiring use of the Plot class.

import ezcharts as ezc
from ezcharts.components.reports.comp import ComponentReport
import seaborn as sns

tips = sns.load_dataset("tips")

plt = ezc.scatterplot(data=tips, x="total_bill", y="tip", hue="size")
ComponentReport.from_plot(plt, "tmp.html")

Layout

The layout functionality of ezCharts uses bootstrap scripting and styling by default, but permits any level of customisation. Snippets provide simple re-usable bits of HTML that are pre-styled, such as tabs and tables.

import ezcharts as ezc
from ezcharts.components.ezchart import EZChart
from ezcharts.layout.snippets import DataTable, Tabs

from ezcharts.components.reports.labs import BasicReport

from dominate.tags import p
import seaborn as sns

tips = sns.load_dataset("tips")
tips["size"] = tips["size"].astype(str)

report = BasicReport("test report")

plt = ezc.scatterplot(data=tips, x="total_bill", y="tip", hue="size")

with report.add_section("test section", "test"):
    tabs = Tabs()
    with tabs.add_tab("First tab"):
        EZChart(plt)
    with tabs.add_tab("Second tab"):
        p("Some text.")
        DataTable.from_pandas(tips)

report.write("tmp.html")

Note that EZChart() is required to add the plot to the enclosing dominate context.

Debug mode

Decorator for plotting a message when the plot fails can be skipped by setting the env variable EZCHARTS_DEBUG=1.

Components

Components provide higher level application-specific layouts (a table for Nextclade results, for instance) that may also include charts and light data processing capabilities (e.g. to parse and visualise fastcat output).

For a comprehensive showcase of current capabilities, please have a look at ezcharts/demo.py.

Contributing

The aim is to slowly build out both the seaborn-like API and the components library with functionality required.

Much of the seaborn data analysis code can be reused. Function stubs have been added according to the v0.11.2 documentation. The seaborn requirement is however pinned to v0.12.0b2. In implementing a plotting function the 0.12.0 series should be followed.