stats.py

import pandas as pd
import numpy as np
import bokeh
import itertools

from bokeh.models import Select, HoverTool, NumeralTickFormatter, ColumnDataSource, Div
from bokeh.plotting import curdoc, figure, show, output_file, save
from bokeh.io import output_notebook
# from bokeh.models.widgets import CheckboxGroup
from bokeh.layouts import widgetbox, row, column

from collections import OrderedDict

from bokeh.layouts import layout
from bokeh.models.widgets import Tabs, Panel
from bokeh.io import curdoc
from bokeh.embed import file_html
from bokeh.resources import CDN
from os.path import dirname, join

##### Load the data file

file = "stats.tsv"
df = pd.read_csv(file, sep="\t")

df = df.drop(['Assembly.quality', 'Source', 'Release raw and assemblies in data note'], axis=1)

columns = sorted(df.columns)
discrete = [x for x in columns if df[x].dtype == object]
continuous = [x for x in columns if x not in discrete]

TOOLS = "pan,wheel_zoom,box_zoom,reset,hover,previewsave"
SIZES = list(range(6, 22, 3))
COLORS = ['#175676', '#A4036F', '#419D78', '#E0A458']
N_SIZES = len(SIZES)
N_COLORS = len(COLORS)


def create_figure():
    """
    Construct and return a scatter plot from a Pandas dataframe.
    :return: pandas plot
    """

    """
    Define lists of the values that belong to the columns chosen as x and y values, then
    set their titles.
    """
    xs = df[x.value].values
    ys = df[y.value].values
    x_title = x.value.title()
    y_title = y.value.title()

    """
    Declare a dictionary and set its key:value pairs according to the values that appear in the
    chosen data for the x and y axes.
    """
    kw = dict()
    if x.value in discrete:
        kw['x_range'] = sorted(set(xs))
    if y.value in discrete:
        kw['y_range'] = sorted(set(ys))
    kw['title'] = "%s vs %s" % (x_title, y_title)

    """
    Declare a plot and set its axis labels.
    """
    p = figure(plot_width=1200, plot_height=550, **kw)
    p.xaxis.axis_label = x_title
    p.yaxis.axis_label = y_title

    """
    If the x-values are discrete, rotate the text for these values 45 degrees counterclockwise.
    """
    if x.value in discrete:
        p.xaxis.major_label_orientation = pd.np.pi / 4

    sz = 9
    if size.value != 'None':
        if len(set(df[size.value])) > N_SIZES:
            groups = pd.qcut(df[size.value].values, N_SIZES, duplicates='drop')
        else:
            groups = pd.Categorical(df[size.value])
        sz = [SIZES[xx] for xx in groups.codes]

    c = "#0B3142"
    if color.value != 'None':
        if len(set(df[color.value])) > N_COLORS:
            groups = pd.qcut(df[color.value].values, N_COLORS, duplicates='drop')
            # groups = pd.Categorical(df[color.value])
        else:
            groups = pd.Categorical(df[color.value])
        c = [COLORS[xx] for xx in groups.codes]

    """
    Represent the data points on the plot as circles.
    """
    p.circle(x=xs, y=ys, color=c, size=sz, line_color="white", alpha=0.8, hover_color='white', hover_alpha=0.5)
    #    p.circle(x=xs, y=ys, line_color="white", alpha=0.6, hover_color='white', hover_alpha=0.5)

    return p


def update(attr, old, new):
    layout.children[1] = create_figure()


x = Select(title='X-Axis', value='Latin_name', options=columns)
x.on_change('value', update)
y = Select(title='Y-Axis', value='Contig_N50..bp.', options=columns)
y.on_change('value', update)

size = Select(title='Size', value='None', options=['None'] + continuous)
size.on_change('value', update)

color = Select(title='Color', value='Source2', options=['None', 'Source2'])
color.on_change('value', update)

controls = row([x, y, color, size], width=400)
layout = column(controls, create_figure())

curdoc().add_root(layout)
# output_file("genome_statistics.html")
# save(layout)
show(layout)
print("x:\n", x)
print("\n\nx.value:\n", x.value)
print("\n\ndf[x.value]\n", df[x.value])
print("\n\ndf[x.value]\n", df[x.value].values)
print("\n\ny:\n", y)
print("\n\ny.value:\n", y.value)
print("\n\ndf[y.value]\n", df[y.value])
print("\n\ndf[y.value]\n", df[y.value].values)