qasim - 1 year ago 94
Python Question

# how to get a density/probability plot using matplotlib

I was trying to get the top-left and top-right panels of the figure 7 in this paper. I got the top-left panel of the figure 7 but couldn't get its top-right panel. The density part of my code generates the green and blue lines in output, which are not correct. How can I get the probability part (top-right panel of the figure 7) and combine it with my figure?

A small part of the input file (input.txt):

0.0000000 0.0000474 0.0000393
400.0000000 0.1775423 0.1091695
800.0000000 0.2363394 0.1158220
1200.0000000 0.2146373 0.1323802
1600.0000000 0.2629943 0.1379013
2000.0000000 0.2353280 0.1205457
2400.0000000 0.2548243 0.1285356
2800.0000000 0.2507923 0.1243078
3200.0000000 0.3038598 0.1328937
3600.0000000 0.2438334 0.1171351
4000.0000000 0.2399136 0.1386342
4400.0000000 0.2263989 0.1232137
4800.0000000 0.2036292 0.1274123
5200.0000000 0.2136007 0.1262307
5600.0000000 0.2685070 0.1408818
6000.0000000 0.2805652 0.1222442
6400.0000000 0.2328329 0.1256370
6800.0000000 0.2660308 0.1135865
7200.0000000 0.2446094 0.1089109
7600.0000000 0.2729914 0.1254719
8000.0000000 0.3119634 0.1378875
8400.0000000 0.3347659 0.1309574
8800.0000000 0.3206002 0.1289072
9200.0000000 0.2670084 0.1275363
9600.0000000 0.2712551 0.1324258
10000.0000000 0.2453061 0.1368878

Code:

#!/usr/bin/python
import numpy as np
import pylab as plot
import matplotlib.pyplot as plt
import numpy, scipy, pylab, random
from matplotlib.ticker import MultipleLocator
import matplotlib as mpl
from matplotlib.ticker import MaxNLocator
from scipy import stats

with open("input.xvg", "r") as f:
x=[]
y1=[]
y2=[]
for line in f:
if not line.strip() or line.startswith('@') or line.startswith('#'): continue
row = line.split()
x.append(float(row[0])*0.001)
y1.append(float(row[1]))
y2.append(float(row[2]))

fig = plt.figure(figsize=(3.2,2.2), dpi=300)
ax = plt.subplot(111)

plt.xlim(0, 1000)
plt.ylim(0, 0.7)
ax.xaxis.set_major_locator(MaxNLocator(10))
ax.yaxis.set_major_locator(MaxNLocator(7))
ax.xaxis.set_minor_locator(MultipleLocator(50))
ax.yaxis.set_minor_locator(MultipleLocator(0.05))

plt.plot(x, y1, 'orange', label='A', linewidth=0.5)
plt.plot(x, y2, 'black', label='B', linewidth=0.5)

plt.xlabel('Time (ns)', fontsize=8)
plt.ylabel('RMSD (nm)', fontsize=8)

for axis in ['top','bottom','left','right']:
ax.spines[axis].set_linewidth(0.5)

plt.tick_params(axis='both', which='major', labelsize=7)
plt.tick_params(axis='both', which='minor', labelsize=0)

#for the density part
density1 = stats.kde.gaussian_kde(y1)
density2 = stats.kde.gaussian_kde(y2)
plt.plot(x, density1(y1))
plt.plot(x, density2(y2))

plt.savefig("output.png", dpi=300)

Output:

The first thing that you have to notice is that the two plots are on separate axes though they share the same y-axis. It would be much more easier to solve the problem without the extra formatting first, then you can apply your special formatting.

#!/usr/bin/python
import numpy as np
import pylab as plot
import matplotlib.pyplot as plt
import numpy, scipy, pylab, random
from matplotlib.ticker import MultipleLocator
import matplotlib as mpl
from matplotlib.ticker import MaxNLocator
from scipy import stats

with open("input.txt", "r") as f:
x=[]
y1=[]
y2=[]
for line in f:
if not line.strip() or line.startswith('@') or line.startswith('#'): continue
row = line.split()
x.append(float(row[0])*0.001)
y1.append(float(row[1]))
y2.append(float(row[2]))

fig, (ax1, ax2) =plt.subplots(1, 2, sharey=True)

ax1.axis([0, 10, 0, 0.7])

ax1.plot(x, y1, 'orange', label='A',  linewidth=1)
ax1.plot(x, y2, 'black', label='B',  linewidth=1)

#for the density part

density1 = stats.kde.gaussian_kde(y1)
density2 = stats.kde.gaussian_kde(y2)

# plot the pdf for the full range of y-axis
y_range = np.linspace(0, 0.7, 100)
ax2.plot(density1(y_range), y_range, 'orange')
ax2.plot(density2(y_range), y_range, 'black')

# display y-axis tick on the right
ax2.yaxis.tick_right()
# remove the spacing between the two axes

# deal with the overlaping x-axis label at the center
# you can remove the label corresponding to the last element of the frist axis
xticks = ax1.xaxis.get_major_ticks()
xticks[-1].label1.set_visible(False)

# modifying the number of y ticks
ax2.yaxis.set_major_locator(MaxNLocator(4.0))
ax2.yaxis.set_minor_locator(MultipleLocator(0.1))

plt.savefig("output.png", dpi=300)

The result looks like

The rest is mere formatting.

Recommended from our users: Dynamic Network Monitoring from WhatsUp Gold from IPSwitch. Free Download