#!/usr/bin/env python
# -*- coding: utf-8 -*-
# pauvre
# Copyright (c) 2016-2020 Darrin T. Schultz.
#
# This file is part of pauvre.
#
# pauvre is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# pauvre is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with pauvre. If not, see <http://www.gnu.org/licenses/>.
import ast
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import matplotlib.patches as mplpatches
from matplotlib.colors import LinearSegmentedColormap
import numpy as np
import pandas as pd
import os.path as opath
from sys import stderr
from pauvre.functions import parse_fastq_length_meanqual, print_images, filter_fastq_length_meanqual
from pauvre.stats import stats
import pauvre.rcparams as rc
import logging
# logging
logger = logging.getLogger('pauvre')
def generate_panel(panel_left, panel_bottom, panel_width, panel_height,
axis_tick_param='both', which_tick_param='both',
bottom_tick_param=True, label_bottom_tick_param=True,
left_tick_param=True, label_left_tick_param=True,
right_tick_param=False, label_right_tick_param=False,
top_tick_param=False, label_top_tick_param=False):
"""
Setting default panel tick parameters. Some of these are the defaults
for matplotlib anyway, but specifying them for readability. Here are
options and defaults for the parameters used below:
axis : {'x', 'y', 'both'}; which axis to modify; default = 'both'
which : {'major', 'minor', 'both'}; which ticks to modify;
default = 'major'
bottom, top, left, right : bool or {True, False}; ticks on or off;
labelbottom, labeltop, labelleft, labelright : bool or {True, False}
"""
# create the panel
panel_rectangle = [panel_left, panel_bottom, panel_width, panel_height]
panel = plt.axes(panel_rectangle)
# Set tick parameters
panel.tick_params(axis=axis_tick_param,
which=which_tick_param,
bottom=bottom_tick_param,
labelbottom=label_bottom_tick_param,
left=left_tick_param,
labelleft=label_left_tick_param,
right=right_tick_param,
labelright=label_right_tick_param,
top=top_tick_param,
labeltop=label_top_tick_param)
return panel
def _generate_histogram_bin_patches(panel, bins, bin_values, horizontal=True):
"""This helper method generates the histogram that is added to the panel.
In this case, horizontal = True applies to the mean quality histogram.
So, horizontal = False only applies to the length histogram.
"""
l_width = 0.0
f_color = (0.5, 0.5, 0.5)
e_color = (0, 0, 0)
if horizontal:
for step in np.arange(0, len(bin_values), 1):
left = bins[step]
bottom = 0
width = bins[step + 1] - bins[step]
height = bin_values[step]
hist_rectangle = mplpatches.Rectangle((left, bottom), width, height,
linewidth=l_width,
facecolor=f_color,
edgecolor=e_color)
panel.add_patch(hist_rectangle)
else:
for step in np.arange(0, len(bin_values), 1):
left = 0
bottom = bins[step]
width = bin_values[step]
height = bins[step + 1] - bins[step]
hist_rectangle = mplpatches.Rectangle((left, bottom), width, height,
linewidth=l_width,
facecolor=f_color,
edgecolor=e_color)
panel.add_patch(hist_rectangle)
def generate_histogram(panel, data_list, max_plot_length, min_plot_length,
bin_interval, hist_horizontal=True,
left_spine=True, bottom_spine=True,
top_spine=False, right_spine=False, x_label=None,
y_label=None):
bins = np.arange(0, max_plot_length, bin_interval)
bin_values, bins2 = np.histogram(data_list, bins)
# hist_horizontal is used for quality
if hist_horizontal:
panel.set_xlim([min_plot_length, max_plot_length])
panel.set_ylim([0, max(bin_values * 1.1)])
# and hist_horizontal == Fale is for read length
else:
panel.set_xlim([0, max(bin_values * 1.1)])
panel.set_ylim([min_plot_length, max_plot_length])
# Generate histogram bin patches, depending on whether we're plotting
# vertically or horizontally
_generate_histogram_bin_patches(panel, bins, bin_values, hist_horizontal)
panel.spines['left'].set_visible(left_spine)
panel.spines['bottom'].set_visible(bottom_spine)
panel.spines['top'].set_visible(top_spine)
panel.spines['right'].set_visible(right_spine)
if y_label is not None:
panel.set_ylabel(y_label)
if x_label is not None:
panel.set_xlabel(x_label)
def generate_heat_map(panel, data_frame, min_plot_length, min_plot_qual,
max_plot_length, max_plot_qual, color, **kwargs):
panel.set_xlim([min_plot_qual, max_plot_qual])
panel.set_ylim([min_plot_length, max_plot_length])
if kwargs["kmerdf"]:
hex_this = data_frame.query('length<{} and numks<{}'.format(
max_plot_length, max_plot_qual))
# This single line controls plotting the hex bins in the panel
hex_vals = panel.hexbin(hex_this['numks'], hex_this['length'], gridsize=int(np.ceil(max_plot_qual/2)),
linewidths=0.0, cmap=color)
else:
hex_this = data_frame.query('length<{} and meanQual<{}'.format(
max_plot_length, max_plot_qual))
# This single line controls plotting the hex bins in the panel
hex_vals = panel.hexbin(hex_this['meanQual'], hex_this['length'], gridsize=49,
linewidths=0.0, cmap=color)
for each in panel.spines:
panel.spines[each].set_visible(False)
counts = hex_vals.get_array()
return counts
def generate_legend(panel, counts, color):
# completely custom for more control
panel.set_xlim([0, 1])
panel.set_ylim([0, 1000])
panel.set_yticks([int(x) for x in np.linspace(0, 1000, 6)])
panel.set_yticklabels([int(x) for x in np.linspace(0, max(counts), 6)])
for i in np.arange(0, 1001, 1):
rgba = color(i / 1001)
alpha = rgba[-1]
facec = rgba[0:3]
hist_rectangle = mplpatches.Rectangle((0, i), 1, 1,
linewidth=0.0,
facecolor=facec,
edgecolor=(0, 0, 0),
alpha=alpha)
panel.add_patch(hist_rectangle)
panel.spines['top'].set_visible(False)
panel.spines['left'].set_visible(False)
panel.spines['bottom'].set_visible(False)
panel.yaxis.set_label_position("right")
panel.set_ylabel('Number of Reads')
def margin_plot(df, **kwargs):
rc.update_rcParams()
# 250, 231, 34 light yellow
# 67, 1, 85
# R=np.linspace(65/255,1,101)
# G=np.linspace(0/255, 231/255, 101)
# B=np.linspace(85/255, 34/255, 101)
# R=65/255, G=0/255, B=85/255
Rf = 65 / 255
Bf = 85 / 255
pdict = {'red': ((0.0, Rf, Rf),
(1.0, Rf, Rf)),
'green': ((0.0, 0.0, 0.0),
(1.0, 0.0, 0.0)),
'blue': ((0.0, Bf, Bf),
(1.0, Bf, Bf)),
'alpha': ((0.0, 0.0, 0.0),
(1.0, 1.0, 1.0))
}
# Now we will use this example to illustrate 3 ways of
# handling custom colormaps.
# First, the most direct and explicit:
purple1 = LinearSegmentedColormap('Purple1', pdict)
# set the figure dimensions
fig_width = 1.61 * 3
fig_height = 1 * 3
fig = plt.figure(figsize=(fig_width, fig_height))
# set the panel dimensions
heat_map_panel_width = fig_width * 0.5
heat_map_panel_height = heat_map_panel_width * 0.62
# find the margins to center the panel in figure
fig_left_margin = fig_bottom_margin = (1 / 6)
# lengthPanel
length_panel_width = (1 / 8)
# the color Bar parameters
legend_panel_width = (1 / 24)
# define padding
h_padding = 0.02
v_padding = 0.05
# Set whether to include y-axes in histograms
if kwargs["Y_AXES"]:
length_bottom_spine = True
length_bottom_tick = False
length_bottom_label = True
qual_left_spine = True
qual_left_tick = True
qual_left_label = True
qual_y_label = 'Count'
else:
length_bottom_spine = False
length_bottom_tick = False
length_bottom_label = False
qual_left_spine = False
qual_left_tick = False
qual_left_label = False
qual_y_label = None
panels = []
# Quality histogram panel
qual_panel_left = fig_left_margin + length_panel_width + h_padding
qual_panel_width = heat_map_panel_width / fig_width
qual_panel_height = length_panel_width * fig_width / fig_height
qual_panel = generate_panel(qual_panel_left,
fig_bottom_margin,
qual_panel_width,
qual_panel_height,
left_tick_param=qual_left_tick,
label_left_tick_param=qual_left_label)
panels.append(qual_panel)
# Length histogram panel
length_panel_bottom = fig_bottom_margin + qual_panel_height + v_padding
length_panel_height = heat_map_panel_height / fig_height
length_panel = generate_panel(fig_left_margin,
length_panel_bottom,
length_panel_width,
length_panel_height,
bottom_tick_param=length_bottom_tick,
label_bottom_tick_param=length_bottom_label)
panels.append(length_panel)
# Heat map panel
heat_map_panel_left = fig_left_margin + length_panel_width + h_padding
heat_map_panel_bottom = fig_bottom_margin + qual_panel_height + v_padding
heat_map_panel = generate_panel(heat_map_panel_left,
heat_map_panel_bottom,
heat_map_panel_width / fig_width,
heat_map_panel_height / fig_height,
bottom_tick_param=False,
label_bottom_tick_param=False,
left_tick_param=False,
label_left_tick_param=False)
panels.append(heat_map_panel)
heat_map_panel.set_title(kwargs["title"])
# Legend panel
legend_panel_left = fig_left_margin + length_panel_width + \
(heat_map_panel_width / fig_width) + (h_padding * 2)
legend_panel_bottom = fig_bottom_margin + qual_panel_height + v_padding
legend_panel_height = heat_map_panel_height / fig_height
legend_panel = generate_panel(legend_panel_left, legend_panel_bottom,
legend_panel_width, legend_panel_height,
bottom_tick_param = False,
label_bottom_tick_param = False,
left_tick_param = False,
label_left_tick_param = False,
right_tick_param = True,
label_right_tick_param = True)
panels.append(legend_panel)
# Set min and max viewing window for length
if kwargs["plot_maxlen"]:
max_plot_length = kwargs["plot_maxlen"]
else:
max_plot_length = int(np.percentile(df['length'], 99))
min_plot_length = kwargs["plot_minlen"]
# Set length bin sizes
if kwargs["lengthbin"]:
length_bin_interval = kwargs["lengthbin"]
else:
# Dividing by 80 is based on what looks good from experience
length_bin_interval = int(max_plot_length / 80)
# length_bins = np.arange(0, max_plot_length, length_bin_interval)
# Set max and min viewing window for quality
if kwargs["plot_maxqual"]:
max_plot_qual = kwargs["plot_maxqual"]
elif kwargs["kmerdf"]:
max_plot_qual = np.ceil(df["numks"].median() * 2)
else:
max_plot_qual = max(np.ceil(df['meanQual']))
min_plot_qual = kwargs["plot_minqual"]
# Set qual bin sizes
if kwargs["qualbin"]:
qual_bin_interval = kwargs["qualbin"]
elif kwargs["kmerdf"]:
qual_bin_interval = 1
else:
# again, this is just based on what looks good from experience
qual_bin_interval = max_plot_qual / 85
qual_bins = np.arange(0, max_plot_qual, qual_bin_interval)
# Generate length histogram
generate_histogram(length_panel, df['length'], max_plot_length, min_plot_length,
length_bin_interval, hist_horizontal=False,
y_label='Read Length', bottom_spine=length_bottom_spine)
# Generate quality histogram
if kwargs["kmerdf"]:
generate_histogram(qual_panel, df['numks'], max_plot_qual, min_plot_qual,
qual_bin_interval, x_label='number of kmers',
y_label=qual_y_label, left_spine=qual_left_spine)
else:
generate_histogram(qual_panel, df['meanQual'], max_plot_qual, min_plot_qual,
qual_bin_interval, x_label='Phred Quality',
y_label=qual_y_label, left_spine=qual_left_spine)
# Generate heat map
counts = generate_heat_map(heat_map_panel, df, min_plot_length, min_plot_qual,
max_plot_length, max_plot_qual, purple1, kmerdf = kwargs["kmerdf"])
# Generate legend
generate_legend(legend_panel, counts, purple1)
# inform the user of the plotting window if not quiet mode
if not kwargs["QUIET"]:
print("""plotting in the following window:
{0} <= Q-score (x-axis) <= {1}
{2} <= length (y-axis) <= {3}""".format(
min_plot_qual, max_plot_qual, min_plot_length, max_plot_length),
file=stderr)
# Print image(s)
if kwargs["BASENAME"] is None:
file_base = opath.splitext(opath.basename(kwargs["fastq"]))[0]
else:
file_base = kwargs["BASENAME"]
print_images(
file_base,
image_formats=kwargs["fileform"],
dpi=kwargs["dpi"],
no_timestamp = kwargs["no_timestamp"],
transparent=kwargs["TRANSPARENT"])
def run(args):
if args.kmerdf:
df = pd.read_csv(args.kmerdf, header='infer', sep='\t')
df["kmers"] = df["kmers"].apply(ast.literal_eval)
else:
df = parse_fastq_length_meanqual(args.fastq)
df = filter_fastq_length_meanqual(df, args.filt_minlen, args.filt_maxlen,
args.filt_minqual, args.filt_maxqual)
stats(df, args.fastq, False)
margin_plot(df=df.dropna(), **vars(args))