comparing p-values from independence tests between R and scipy

pval_compare.py

import numpy as np
import scipy.stats as stats
import rpy
import matplotlib.pyplot as plt

spy_p_paired = []
spy_p_nopair = []
rpy_p_paired = []
rpy_p_nopair = []
for n in range(1000):
    # fake data
    y1 = np.random.lognormal(mean=0.57, sigma=1.35, size=37)
    y2 = np.random.lognormal(mean=0.57, sigma=1.35, size=45)
    x = np.random.lognormal(mean=0.51, sigma=1.15, size=37)

    # indep. tests in R
    rpy_paired = rpy.r.wilcox_test(x, y1, paired=True)
    rpy_notpaired = rpy.r.wilcox_test(x, y2, paired=False)

    # inped. tests in scipy
    U, Up = stats.mannwhitneyu(x,y2)
    W, Wp = stats.wilcoxon(x, y1)


    spy_p_paired.append(Wp)
    spy_p_nopair.append(Up*2)
    rpy_p_paired.append(rpy_paired['p.value'])
    rpy_p_nopair.append(rpy_notpaired['p.value'])

fig, ax = plt.subplots()
ax.loglog(spy_p_paired, rpy_p_paired, 'k.')
ax.xaxis.grid(False, which='minor')
ax.yaxis.grid(False, which='minor')
ax.set_xlabel('Scipy p-val')
ax.set_ylabel('R p-val')
ax.set_xlim([0.0001, 1])
ax.set_ylim([0.0001, 1])
ax.set_aspect('equal')
plt.show()

fig, ax = plt.subplots()
ax.loglog(spy_p_nopair, rpy_p_nopair, 'k.')
ax.xaxis.grid(False, which='minor')
ax.yaxis.grid(False, which='minor')
ax.set_xlabel('Scipy p-val')
ax.set_ylabel('R p-val')
ax.set_xlim([0.0001, 1])
ax.set_ylim([0.0001, 1])
ax.set_aspect('equal')
plt.show()