def plot():
print('Plotando...')
# ###### PLOT ALL NUMERIC!
data.dropna(inplace=True)
data.drop('same_strand', 1, inplace=True)
data = data.infer_objects()
data = data.select_dtypes(exclude=['object'])
pd.plotting.scatter_matrix(data.drop(['end', 'start'], 1))
data.plot.scatter('start', 'transcription', alpha=.2)
data.plot.scatter('distance', 'transcription', alpha=.2)
plot_box(data[data.distance > 0], 'distance', 'transcription', bins=50)
genome_map = data.sort_values('start')[['start', 'transcription']]
genome_map['start'] = pd.cut(genome_map.start, int(1e4))
genome_map = genome_map.groupby('start').sum()
le = len(genome_map.transcription)
plt.figure()
plt.pcolor(
genome_map.transcription.values.reshape(int(le**.5), int(le**.5)))
plt.colorbar()
plt.figure()
plt.pcolor(data.corr(method='spearman'))
plt.figure()
data.transcription.hist()
save_all_figs()
def plot(counts):
hist = counts.value_counts()
print(hist)
soma = hist.sum()
print('soma:', soma, 'fração de 0 repetições:',
hist[0]/soma, 'fração com até 1 rep.', hist[0:2].sum()/soma)
for log in (False, True):
print(f'Construindo histograma, log={log}')
plt.figure(figsize=(11, 4.8))
# ax = counts.hist(bins=200, figsize=(4, 4.8))
plt.bar(hist.index, hist, log=log)
plt.xlabel('Número de repetições')
plt.ylabel('Número de sequências')
plt.title('Quantidade de cópias repetidas')
print("Histograma salvo.")
save_all_figs()
def main():
#print('Quantidades das populações')
#[print(name, s) for name, s in u.get_subsets(d).items()]
for corr_transcr in ('transcription', 'complement_transcription',
'gene_transcription', 'gene_complement_transcription',
'correlation', 'complement_correlation',
'motherlength', 'repetitions'):
u.print_header(corr_transcr)
# D = d[d[corr_transcr] != 0].dropna(subset=[corr_transcr]) # drop zeros
D = d.dropna(subset=[corr_transcr])
# print(f'Comprimento total: {d.shape[0]} | Comprimento coluna: {D.shape[0]}')
if 'gene' in corr_transcr:
D = D.sort_values('distance') # Keep only closest copy to each G
D = D.drop_duplicates(subset='neighbor_gene')
subsets = get_subsets(D, corr_transcr)
compare(subsets, corr_transcr)
if show_flag:
plt.show()
else:
save_all_figs()
# olap_mask = (max(mask.start - hstart, 0),
# min(mask.end - hstart, 1000))
# for i in range(*olap_mask):
# mask_count[i] += 1
# # plt.fill_between(olap_mask, *axesy, alpha=.1)
# count += 1
#
# print(irow, '/', n_heads, 'plotted:', count)
#
# except KeyboardInterrupt:
# pass
#
# return mask_count
# NORMALIZE
complete_counts = [c / max(complete_counts) for c in complete_counts]
#mask_count = [c/max(mask_count) for c in mask_count]
plt.figure(figsize=(9, 4.8))
plt.plot(complete_counts, label='Read count')
#plt.plot(mask_count, label='Masked count')
plt.title('Perfil geral de transcrição das sequências sonda')
plt.xlabel('Distância ao início da sonda (pb)')
plt.ylabel('Contagem de reads normalizada')
if show_flag:
plt.show()
save_all_figs()
print('Pronto.')
def main(func):
counts_sum = func(counts)
counts_sum = counts_sum.iloc[1:]
genes_data = counts_sum[counts_sum.index.str.startswith('Smp')]
genes_data = genes_data.iloc[~genes_data.index.str.endswith('complement')]
with_perere = head_data.gene_transcription.dropna().drop_duplicates()
lone_genes = genes_data.drop(
head_data.neighbor_gene.dropna().unique()).dropna()
# print(
# with_perere,
# lone_genes,
# genes_data
# )
print(head_data.neighbor_gene.duplicated().sum())
plt.figure(figsize=(6, 10))
u.multibox_compare(
(with_perere, lone_genes, genes_data),
('Com Perere-3 vizinho', 'Sem Perere-3 vizinho', 'Total'),
margin=3)
if __name__ == '__main__':
for func in (pd.DataFrame.sum, pd.DataFrame.max):
main(counts)
u.save_all_figs()
def main():
head_data = pd.read_table(pardir/'genome_annotation/all_together_now.tsv')
# # Very important to drop NaN's! (we use stuff like data[data.relative_position != 'olap'])
# This keeps only heads with neighbor genes.
head_data = head_data.dropna().reset_index(drop=True)
# ################# CORRELATION ########################
print("TABELA DE CORRELAÇÕES DE SPEARMAN")
print(head_data[['transcription', 'correlation', 'distance']].corr(method='spearman'))
# #################### SPLITS ##########################
# drop overlapped
head_data = head_data[head_data.relative_position != 'olap']
# ##### SAME/DIFF STRAND
same_strand = head_data[head_data.same_strand]
diff_strand = head_data.drop(same_strand.index)
# ##### UP/DOWN-STREAM: -----> -->
# downstream = same_strand[(same_strand.strand == '+') &
# (same_strand.relative_position == 'dir')]
# downstream = downstream.append(same_strand[(same_strand.strand == '-') &
# (same_strand.relative_position == 'esq')])
# upstream = same_strand[(same_strand.strand == '+') &
# (same_strand.relative_position == 'esq')]
# upstream = upstream.append(same_strand[(same_strand.strand == '-') &
# (same_strand.relative_position == 'dir')])
# ##==============================================================
downstream = head_data[(head_data.strand == '+') &
(head_data.relative_position == 'dir')]
downstream = downstream.append(head_data[(head_data.strand == '-') &
(head_data.relative_position == 'esq')])
upstream = head_data.drop(downstream.index)
# ################# Wilcoxon #####################
thresholds = (1e3, 1e4, 2e4, upstream.append(downstream).distance.max())
abc = ('a) ', 'b) ', 'c) ')
fig, axs = plt.subplots(1, len(thresholds), figsize=(11, 4.8))
for ax in axs:
ax.get_xaxis().set_ticks([])
ax.get_yaxis().set_ticks([])
ax = fig.add_subplot(111, frameon=False)
ax.grid(False)
plt.tick_params(labelcolor='none',
top=False, bottom=False, left=False, right=False)
plt.ylabel('Correlação transcricional com o gene vizinho')
for i, thresh in enumerate(thresholds):
# Data selection
a, b = [p.loc[p.distance <= thresh].correlation
for p in [downstream, upstream]]
pvalue = mannwhitneyu(a, b).pvalue
plabel = f'p-valor:\n{pvalue:.5f}'
label = f"Distância < {thresh:.0f}"
print(label, pvalue, 'Medians:', a.median(), b.median())
fig.add_subplot(1, len(thresholds), i + 1, frameon=False)
plt.title(label)
boxplot([a, b])
plt.annotate(plabel, (.5, .2), xycoords='axes fraction', ha='center')
plt.xticks([0, 1], labels=['Downstream', 'Upstream'])
plt.tight_layout()
# ======================= P VS. DIS ==========================
plt.figure(figsize=(11, 4.8))
plt.subplot(211)
xdistances = range(100, int(upstream.distance.max()), 100)
ypvalues = [mannwhitneyu(downstream.loc[downstream.distance <= thresh].correlation,
upstream.loc[upstream.distance <= thresh].correlation).pvalue
for thresh in xdistances]
updataamounts = [len(upstream.loc[upstream.distance <= thresh]) for thresh in xdistances]
downdataamounts = [len(downstream.loc[downstream.distance <= thresh]) for thresh in xdistances]
plt.plot(xdistances, ypvalues)
plt.semilogx()
plt.ylabel('p-valor')
# ----------------------------------
plt.subplot(212)
plt.plot(xdistances, downdataamounts, label="Downstream")
plt.plot(xdistances, updataamounts, label="Upstream")
plt.legend()
plt.semilogx()
plt.xlabel('Distância ao vizinho (pb)')
plt.ylabel('Quantidade de pontos')
# ===========================================
# Só na faixa selecionada
# plt.figure(dpi=200)
# limits = 1e3, 1e4
# print(i.distance.between(*limits))
# plt.boxplot([[i.correlation.loc[i.distance.between(*limits)]]
# for i in (upstream, downstream)])
# ===========================================
save_all_figs()
