def rep_plot_pca(df,
x=1,
y=2,
s=300,
figsize=None,
dpi=None,
label=True,
cmap=None,
fontsize=12):
l1, l2 = f"pc{int(x)}", f"pc{int(y)}"
pc1, pc2 = df[l1], df[l2]
gen, rep = df['gen'], df['rep']
plt.scatter(pc1, pc2, c=integerize(rep), s=s, zorder=2, cmap=cmap)
# plot lines between consecutive generations
for rep in rep.unique():
this_rep = df[df['rep'] == rep]
plt.plot(this_rep[l1], this_rep[l2], '--', color='0.8', zorder=1)
# each marker gets a label, of the current replicate
label_df = df[[l1, l2, 'rep', 'gen']]
if label:
for pc1, pc2, rep, gen in label_df.itertuples(index=False):
plt.text(pc1,
pc2,
s=f"{rep}",
horizontalalignment='center',
verticalalignment='center',
fontsize=fontsize)
plt.xlabel(l1.upper())
plt.ylabel(l2.upper())
plt.tight_layout()
return plt
def het_plot(tempfreqs, rep, time):
p = tempfreqs.freqs[rep, time, :]
het = 2 * p * (1 - p)
midpoints = tempfreqs.gintervals.cummulative_midpoint
seqids = tempfreqs.gintervals.seqid
plt.scatter(midpoints, het, c=integerize(seqids))
def correction_diagnostic_plot(diag, figsize=None, color=True):
corr_df, models, xpreds, ypreds = diag
if figsize is not None:
fig, ax = plt.subplots(ncols=2, nrows=2, figsize=figsize)
else:
fig, ax = plt.subplots(ncols=2, nrows=2)
labelx, labely = 0.05, 0.85
before = corr_df[corr_df['correction'] == False]
after = corr_df[corr_df['correction'] == True]
if color:
ax[0, 0].scatter(before['depth'],
before['diag'],
c=integerize(before['seqid']),
s=5)
ax[0, 1].scatter(before['depth'],
after['diag'],
c=integerize(before['seqid']),
s=5)
else:
ax[0, 0].scatter(before['depth'], before['diag'], s=5)
ax[0, 1].scatter(before['depth'], after['diag'], s=5)
ax[0, 0].plot(xpreds[False], ypreds[False][0], 'r-')
ax[0, 1].plot(xpreds[True], ypreds[True][0], 'r-')
ax[0, 0].annotate('before correction',
xy=(labelx, labely),
xycoords='axes fraction')
ax[0, 0].set_ylabel('variance')
ax[0, 0].set_xlabel('average depth per window')
ax[0, 1].set_xlabel('average depth per window')
ax[0, 1].annotate('after correction',
xy=(labelx, labely),
xycoords='axes fraction')
ax[1, 0].plot(xpreds[False], ypreds[False][1], 'r-')
ax[1, 0].annotate('before correction',
xy=(labelx, labely),
xycoords='axes fraction')
ax[1, 0].axhline(y=0, color='99', zorder=1, linestyle='--')
ax[1, 0].set_ylabel('covariance')
if color:
ax[1, 0].scatter(before['depth'],
before['offdiag'],
c=integerize(before['seqid']),
zorder=2,
s=5)
ax[1, 1].scatter(before['depth'],
after['offdiag'],
c=integerize(before['seqid']),
zorder=2,
s=5)
else:
ax[1, 0].scatter(before['depth'], before['offdiag'], zorder=2, s=5)
ax[1, 1].scatter(before['depth'], after['offdiag'], zorder=2, s=5)
ax[1, 1].plot(xpreds[True], ypreds[True][1], 'r-')
ax[1, 1].annotate('after correction',
xy=(labelx, labely),
xycoords='axes fraction')
ax[1, 1].axhline(y=0, color='99', zorder=1, linestyle='--')
ax[1, 0].set_xlabel('average depth per window')
ax[1, 1].set_xlabel('average depth per window')
plt.tight_layout()
return fig, ax