def radar_plot(ax, chrom, FSAMPLE=True):
fig_all_df = cnv_chr_counts.query('Space==@chrom')
# chrlen = ref.query('space==@chrom').max()['end']/1e6/100
chrlen = (100*ref.groupby('space')['end'].max()/ref.groupby('space')['end'].max().sum()).loc[chrom]
# y_all = fig_all_df.set_index('Sample').reindex(index=xlsx_tab['SAMPLE']).fillna(0)['ratio']
y_all = fig_all_df.set_index('Sample').reindex(index=xlsx_tab['SAMPLE']).fillna(0)['Cell']
y_all = y_all/chrlen
if not FSAMPLE:
y_all = y_all.drop('$FSAMPLE')
codes = xlsx_tab.query('SAMPLE!="$FSAMPLE"').replace('M11','M10')['CODE'].values
else:
codes = xlsx_tab['CODE'].values
n_sample = y_all.shape[0]
x = np.linspace(0,2*np.pi,n_sample+1)
fig_df = pd.DataFrame({'x':x[:-1],'y':y_all})
ax.bar(
fig_df['x'],
# fig_df['y'].clip(upper=2),
fig_df['y'].clip(upper=1),
width=0.3,
facecolor='limegreen',
edgecolor='k',
lw=1.5,
alpha=0.9,
zorder=5
)
ax.bar(
# fig_df.query('y>2')['x'],
# (fig_df.query('y>2')['y'])*2/30,
fig_df.query('y>1')['x'],
(fig_df.query('y>1')['y']-1)/10,
width=0.3,
bottom=1,
facecolor='gold',
edgecolor='k',
lw=1.5,
alpha=0.9,
zorder=5
)
# ax.set_rlim(0,4.7)
# ax.set_rlim(-0.7,3.7)
ax.set_rlim(0,3.7)
ax.set_rorigin(-0.7)
# _, r_label = ax.set_rgrids([2,3,4], ['2','15','30'], angle=12, fontsize=smallsize, va='bottom', ha='center')
_, r_label = ax.set_rgrids([1,1.9,2.9], ['1','10','20'], angle=12, fontsize=smallsize, va='bottom', ha='center')
r_label[0].set_color('limegreen')
r_label[1].set_color('darkgoldenrod')
r_label[2].set_color('darkgoldenrod')
_, sample_label = ax.set_thetagrids( np.linspace(0,360,n_sample), codes )
for tobj,angle in zip(sample_label,np.linspace(0,360,n_sample)):
if 'M' in tobj.get_text():
tobj.set_color(two_colors[0])
else:
tobj.set_color(two_colors[1])
tobj.set_fontsize(smallsize)
for tpos,l in zip(x, sample_label):
if 'M' in l.get_text():
r=90
else:
r=270
ax.text(
# tpos, 5.5,
tpos, 4.5,
l.get_text().replace('F','').replace('M',''),
va='center',
ha='center',
fontsize=smallsize,
color=l.get_color(),
rotation=r-tpos*360/np.pi/2
)
ax.set_title(chrom, fontsize=midsize, pad=midsize)
ax.tick_params(grid_color='k', grid_linewidth=1, grid_alpha=1, zorder=0, grid_linestyle='--')
ax.set_theta_zero_location('N')
ax.set_theta_direction(-1)
ax.set_xticklabels([])
return
def female_genome_plot(ax1, ax2):
# f, (ax1, ax2) = plt.subplots(2, 1, sharex=True, figsize=(12,2), gridspec_kw={'height_ratios':[3,1], 'hspace':0.1})
n = pd.concat(
[
cells_tab,
xlsx_tab.set_index('SAMPLE')
], axis=1, sort=False
).query('GENDER=="female"')['n_pf'].sum()
big_cnv_gain =(
sum_df
.loc[big_cnv_idx]
.query('cnvTag=="Gain" & Gender=="female"')
)
big_cnv_loss =(
sum_df
.loc[big_cnv_idx]
.query('cnvTag=="Loss" & Gender=="female"')
)
bin_bed = BedTools.from_dataframe(ref.loc[good_df.index])
gain_bed = BedTools.from_dataframe(big_cnv_gain)
loss_bed = BedTools.from_dataframe(big_cnv_loss)
gain_s = BedTools.intersect(self=bin_bed,b=gain_bed,wa=True,c=True).to_dataframe().set_index(good_df.index)['score']
loss_s = BedTools.intersect(self=bin_bed,b=loss_bed,wa=True,c=True).to_dataframe().set_index(good_df.index)['score']
tmpCleanUp()
ax1.plot(np.arange(female_fig_df.columns.shape[0]), 100*gain_s[female_fig_df.columns].fillna(0)/n, '-', lw=2, color='darkorange', label='Gain')
ax2.plot(np.arange(female_fig_df.columns.shape[0]), 100*gain_s[female_fig_df.columns].fillna(0)/n, '-', lw=2, color='darkorange', label='Gain')
ax1.plot(np.arange(female_fig_df.columns.shape[0]), -100*loss_s[female_fig_df.columns].fillna(0)/n, '-', lw=2, color='deepskyblue', label='Loss')
ax2.plot(np.arange(female_fig_df.columns.shape[0]), -100*loss_s[female_fig_df.columns].fillna(0)/n, '-', lw=2, color='deepskyblue', label='Loss')
# Plot chroms. boundary
g = ref.groupby('space').min()['abs_pos']
for pos in chr_pos:
ax1.plot(np.tile(pos,10), np.linspace(-5, 9, 10), '-', color='black', alpha=0.5, lw=1, zorder=0)
ax2.plot(np.tile(pos,10), np.linspace(-5, 9, 10), '-', color='black', alpha=0.5, lw=1, zorder=0)
ax1.set_xticks([])
ax2.set_xticks([])
ax1.tick_params(axis='x', bottom=False, length=0)
ax2.tick_params(axis='x', bottom=False, length=0)
ax1.tick_params(axis='y', length=5)
ax2.tick_params(axis='y', length=5)
ax1.set_ylim(-0.4, 0.4)
ax2.set_ylim(-4.5, -1.5)
ax1.spines['bottom'].set_visible(False)
ax2.spines['top'].set_visible(False)
# d = 0.01 # how big to make the diagonal lines in axes coordinates
# kwargs = dict(transform=ax1.transAxes, color='k', clip_on=False)
# ax1.plot((-d, +d), (0, 0), **kwargs) # top-left diagonal
# ax1.plot((1 - d, 1 + d), (0, 0), **kwargs) # top-right diagonal
# kwargs.update(transform=ax2.transAxes) # switch to the bottom axes
# ax2.plot((-d, +d), (1, 1), **kwargs) # bottom-left diagonal
# ax2.plot((1 - d, 1 + d), (1, 1), **kwargs) # bottom-right diagonal
ax1.set_ylabel('Freq.\n(%)', fontsize=midsize)
ax1.set_yticks([0.3,0,-0.3])
ax1.set_yticklabels(['0.3','0.0','0.3'], fontsize=smallsize)
ax2.set_yticks([-3])
ax2.set_yticklabels(['3'], fontsize=smallsize)
return
def chr_size_plot(axs, FSAMPLE=True):
# fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(6,4))
axx = axs[0]
ax = axs[1]
if FSAMPLE:
# chr_counts = sum_df.loc[big_cnv_idx].groupby('Space')['Pos'].count()
chr_counts = (
sum_df
.loc[big_cnv_idx]
.groupby(['Space','Cell','cnvTag'])['Pos'].first()
.reset_index()
.groupby('Space')['Cell'].count()
)
else:
chr_counts = (
sum_df
.loc[big_cnv_idx]
.query('Sample!="$FSAMPLE"')
.groupby(['Space','Cell','cnvTag'])['Pos'].first()
.reset_index()
.groupby('Space')['Cell'].count()
)
# x = ref.groupby('space').max().drop(['chrX','chrY'])['end']/1000/1000
# y = chr_counts.drop(['chrX','chrY'], errors='ignore')
x = ref.groupby('space').max().drop(['chrY'])['end']/1000/1000
y = chr_counts.drop(['chrY'], errors='ignore')
ax.plot(
x.drop('chr21'), y.drop('chr21'), '.',
color='black',
markersize=8,
)
ax.plot(
x['chr21'], y['chr21'], '.',
color='red',
markersize=10,
)
axx.plot(
x['chrX'], y['chrX'], '.',
color='red',
markersize=10,
)
for tchr in ref['space'].unique()[:-2]:
a = -5
b = 4
if tchr == 'chr21':
a = 0
b = -13
elif tchr == 'chr17':
a = -5
b = -10
elif tchr == 'chr11':
a = 8
b = 0
elif tchr == 'chr10':
a = -8
b = -3
elif tchr == 'chr2':
a = 0
b = -10
elif tchr == 'chr18':
a = 8
b = -2
elif tchr == 'chr8':
a = 5
b = -3
elif tchr == 'chr12':
pass
elif tchr == 'chr9':
a = 3
b = 4
pass
ax.text(
a + ref.groupby('space').max().loc[tchr,'end']/1000/1000,
b + chr_counts[tchr],
tchr.replace('chr',''),
fontsize=10,
ha='center',
va='center',
)
ax.set_xlim(0,300)
ax.set_xticks([0,100,200,300])
ax.set_ylim(0,180)
ax.set_yticks([0,80,160])
axx.set_xlim(0,300)
axx.set_ylim(280,400)
axx.set_yticks([300,400])
axx.set_xticks([])
axx.set_xticklabels('')
axx.spines['bottom'].set_visible(False)
ax.spines['top' ].set_visible(False)
axx.text(
ref.groupby('space').max().loc['chrX','end']/1000/1000,
chr_counts['chrX']-30,
'X',
fontsize=10,
ha='center',
va='center',
)
# d = 0.015 # how big to make the diagonal lines in axes coordinates
# ax1 = axx
# ax2 = ax
# kwargs = dict(transform=ax1.transAxes, color='k', clip_on=False)
# ax1.plot((-d, +d), (-d, +d), **kwargs)
# kwargs.update(transform=ax2.transAxes)
# ax1.plot((1-d, 1+d), (1-d, 1+d), **kwargs)
# kwargs.update(transform=ax1.transAxes)
# ax2.plot((1-d, 1+d), (-d, +d), **kwargs)
# kwargs.update(transform=ax2.transAxes)
# ax2.plot((-d, +d), (1-d, 1+d), **kwargs)
sns.regplot(
# x.drop('chr21'), y.drop('chr21'),
x.drop(['chr21','chrX']), y.drop(['chr21','chrX']),
ci=95,
ax=ax,
# scatter=False,
scatter_kws={'color':'w','zorder':0, 'alpha':0},
line_kws={'lw':0, 'linestyle':'--', 'color':'k','zorder':0, 'alpha':0.5},
)
# model = LinearRegression(fit_intercept=False).fit( x.drop('chr21').values.reshape(-1,1), y.drop('chr21') )
# r_sq = model.score(x.drop('chr21').values.reshape(-1,1), y.drop('chr21'))
model = LinearRegression(fit_intercept=False).fit( x.drop(['chr21','chrX']).values.reshape(-1,1), y.drop(['chr21','chrX']) )
r_sq = model.score(x.drop(['chr21','chrX']).values.reshape(-1,1), y.drop(['chr21','chrX']))
ax.plot(np.linspace(50,250), model.predict( np.linspace(50,250).reshape((-1, 1)) ), '--', lw=1.5, color='k', zorder=0 )
# texts1 = r'y=$\alpha$x+$\beta$'
# texts2 = r'$R^2$=' + str(np.round(r_sq,2))
texts1 = r'$R^2$=' + str(np.round(r_sq,2))
# texts3 = r'$\alpha$=' + str(np.round(model.coef_[0],4))
# texts4 = r'$\beta$=' + str(np.round(model.intercept_,4))
# texts1 = 'y=ax+b'
# texts2 = 'R2=' + str(np.round(r_sq,2))
# texts3 = 'a=' + str(np.round(model.coef_[0],4))
# texts4 = 'b=' + str(np.round(model.intercept_,4))
ax.text(200, 20, texts1, fontsize=smallsize, va='center' )
# ax.text(20, 135, texts2, fontsize=smallsize, va='center' )
# ax.text(200, 30, texts3, fontsize=smallsize, va='center' )
# ax.text(200, 15, texts4, fontsize=smallsize, va='center' )
ax.tick_params(axis='both', labelsize=smallsize, length=5)
axx.tick_params(axis='both', labelsize=smallsize, length=5)
ax.set_ylabel(' '*8+'# of cells with CNA', fontsize=midsize, labelpad=-5)
ax.set_xlabel('Chromosome size (Mb)', fontsize=midsize)
return
def male_genome_plot(ax, FSAMPLE=True):
# fig, ax = plt.subplots(figsize=(20,4))
n = pd.concat(
[
cells_tab,
xlsx_tab.set_index('SAMPLE')
], axis=1, sort=False
).query('GENDER=="male"')
big_cnv_gain =(
sum_df
.loc[big_cnv_idx]
.query('cnvTag=="Gain" & Gender=="male"')
)
big_cnv_loss =(
sum_df
.loc[big_cnv_idx]
.query('cnvTag=="Loss" & Gender=="male"')
)
if not FSAMPLE:
big_cnv_gain = big_cnv_gain.query('Sample!="$FSAMPLE"')
big_cnv_loss = big_cnv_loss.query('Sample!="$FSAMPLE"')
n = n.drop('$FSAMPLE')['n_pf'].sum()
else:
n = n['n_pf'].sum()
bin_bed = BedTools.from_dataframe(ref.loc[good_df.index])
gain_bed = BedTools.from_dataframe(big_cnv_gain)
loss_bed = BedTools.from_dataframe(big_cnv_loss)
gain_s = BedTools.intersect(self=bin_bed,b=gain_bed,wa=True,c=True).to_dataframe().set_index(good_df.index)['score']
loss_s = BedTools.intersect(self=bin_bed,b=loss_bed,wa=True,c=True).to_dataframe().set_index(good_df.index)['score']
tmpCleanUp()
ax.plot(np.arange(male_fig_df.columns.shape[0]), 100*gain_s[male_fig_df.columns].fillna(0)/n, '-', lw=2, color='darkorange', label='Gain', zorder=1)
ax.plot(np.arange(male_fig_df.columns.shape[0]), -100*loss_s[male_fig_df.columns].fillna(0)/n, '-', lw=2, color='deepskyblue', label='Loss', zorder=1)
# Plot chroms. boundary
g = ref.groupby('space').min()['abs_pos']
for pos in chr_pos:
ax.plot(np.tile(pos,10), np.linspace(-3, 9, 10), '-', color='black', alpha=0.5, lw=1, zorder=0)
ax.set_ylabel('Freq.\n(%)', fontsize=midsize)
ax.set_xticks( pd.Series(chr_pos).rolling(2).mean()[1:] )
ax.set_xticklabels(
ref['space'].drop_duplicates()[:-1].str.replace('chr','').replace('19',' 19').replace('21',' 21').replace('X',' X'),
fontsize=smallsize-4,
ha='center',
rotation=90,
)
ax.xaxis.set_ticks_position('top')
ax.set_ylim(-0.4,0.4)
ax.set_yticks( [-0.3, 0, 0.3] )
ax.set_yticklabels( [0.3, 0 ,0.3], fontsize=smallsize )
ax.tick_params(axis='x', bottom=False, length=0, pad=5)
ax.tick_params(axis='y', length=5)
return
gridspec_kw={'hspace':0.2, 'wspace':0.5},
)
axs = axs.flatten()
for n,chrom in enumerate(ref['space'].unique()[:-1]):
radar_plot(axs[n],chrom, FSAMPLE=FSAMPLE)
axs[-2].axis('off')
axs[-1].axis('off')
fig.savefig('A4_SI_Radar.pdf', frameon=False, transparent=True, bbox_inches='tight')
plt.show()
# +
chrlen = (100*ref.groupby('space')['end'].max()/ref.groupby('space')['end'].max().sum())
cnv_chr_counts['norm_Cell'] = (cnv_chr_counts.set_index('Space')['Cell']/chrlen).dropna().values
df = cnv_chr_counts.pivot(index='Space',columns='Sample', values='norm_Cell').fillna(0).loc[
ref['space'].unique(),
xlsx_tab['SAMPLE'],
].drop('$FSAMPLE',axis=1).drop('chrY')
a4f = 2.5
width_a4f = 0.4
hight_a4f = 0.35
fig, ax = plt.subplots(
figsize=(a4f*width_a4f*8.27,a4f*hight_a4f*11.69),
)
'Space=="chr6" & Sample=="XXX" & chrom_fraction>15 & chrom_fraction<50'
)['Cell'].unique()
lax = None
for i, n in enumerate(np.arange(1, 30, 3)):
# for i,n in enumerate(np.arange(1,33,3)):
ax_cell = fig.add_subplot(gs[n:n + 3, 20:])
showCell(clone_cells[i], give_ax=[ax_cell], alpha=0.5, ms=1)
ax_cell.set_ylabel('')
ax_cell.yaxis.set_ticks_position('right')
ax_cell.tick_params(axis='y', labelsize=smallsize, length=5)
if not lax:
lax = ax_cell
lax.set_xticks(
ref.groupby('space', sort=False)['abs_pos'].min().rolling(2).mean()[1:])
lax.set_xticklabels(
ref['space'].drop_duplicates()[:-1].str.replace('chr', '').replace(
'17',
' 17').replace('19',
' 19').replace('21',
' 21').replace('X', ' X'),
fontsize=smallsize,
ha='center',
rotation=90,
)
lax.xaxis.set_ticks_position('top')
lax.tick_params(axis='x', bottom=False, length=0, pad=5)
lax.set_title('F01 chr6 clonal CNA cells', fontsize=midsize, pad=-5)
fig.subplots_adjust(hspace=10, wspace=0.4)
def ideoPlot(chrom,ax):
'''
Plot ideo for one chromosome.
'''
tag_acen = 1
chr_base = ref.groupby('space').min().loc[chrom,'abs_pos']
rdis = ref.groupby('space').max().loc[chrom,'end']*rf
for idx, row in ideo.set_index('#chrom').loc[chrom].reset_index().iterrows():
start = row['chromStart'] + chr_base
end = row['chromEnd'] + chr_base
gieStain = row['gieStain']
if idx == 0:
ax.set_xlim(start-rdis, ax.get_xlim()[1])
v1 = [
( end, tf ),
( end, bf ),
( start, bf ),
( start-rdis, bf ),
( start-rdis, tf ),
( start, tf ),
( end, tf ),
]
p = PathPatch( Path(v1, codes), facecolor=color_dict[gieStain], edgecolor='black', lw=2, zorder=1 )
ax.add_patch(p)
elif idx == ideo.set_index('#chrom').loc[chrom].shape[0]-1:
ax.set_xlim(ax.get_xlim()[0], end+rdis)
v2 = [
( start, bf ),
( start, tf ),
( end, tf ),
( end+rdis, tf ),
( end+rdis, bf ),
( end, bf ),
( start, bf ),
]
p = PathPatch( Path(v2, codes), facecolor=color_dict[gieStain], edgecolor='black', lw=2, zorder=1 )
ax.add_patch(p)
elif gieStain == 'acen':
if tag_acen == 1:
triangle = [
(start, bf),
(start, tf),
(end, mf)
]
tag_acen = 0
elif tag_acen == 0:
triangle = [
(end, bf),
(end, tf),
(start, mf)
]
tag_acen = -1
else:
print ('Bug at acen!')
continue
p = Polygon( triangle, facecolor=color_dict[gieStain], edgecolor='black', lw=2, zorder=1 )
ax.add_patch(p)
else:
p = Rectangle( (start,bf), end-start, height=(tf-bf), facecolor=color_dict[gieStain], edgecolor='black', lw=2, zorder=1 )
ax.add_patch(p)
ax.spines['top' ].set_visible(False)
ax.spines['bottom'].set_visible(False)
ax.spines['left' ].set_visible(False)
ax.spines['right' ].set_visible(False)
ax.tick_params(bottom=False, length=2, pad=-20)
ax.set_ylim(-4,1)
ax.set_yticks([])
chrom_bins = chrom_dict[chrom]
boundary = max(2, chrom_bins//50)
step = (chrom_bins+boundary)//5
ideo_zero = ref.groupby('space').min().loc[chrom,'abs_pos']
ax.set_xticks(
ideo_zero + np.arange( 0, (chrom_bins+boundary)+step, step )*1e6
)
ax.set_xticklabels(
np.arange( 0, (chrom_bins+boundary)+step, step ),
fontsize=midsize,
)
ax.set_xlim(ideo_zero-boundary*1e6,ideo_zero+(chrom_bins+boundary)*1e6)
return ax
]
np.random.seed(0)
# -
# # Get data
max_alpha = 4
max_pie = 40
# +
good_cells = good_df.xs(key='copy',axis=1,level=1).columns
sample_code_dict = xlsx_tab.set_index('SAMPLE')['CODE'].to_dict()
chrom_dict = (ref.groupby('space').max()['end']//1e6).astype(int).to_dict()
two_colors = sns.xkcd_palette(['red', 'bright blue'])
def name2code(name):
code = []
for i in name.split('_'):
if i in sample_code_dict:
code.append(sample_code_dict[i])
else:
code.append(i)
return '_'.join(code)
# -
