def sort_results(rvcf_input_file_path, taxon_table_file_path, transform,
r_sqr_median_cutoff, stability_cutoff, snp_count, no_tables,
extra_columns):
print('plotting {} SNPs from {}'.format(snp_count, rvcf_input_file_path))
# read the rvcf file and sort by rsq_median
df = pd.read_csv(rvcf_input_file_path, sep='\t')
#print('df.shape: {}'.format(df.shape))
sorted_rsq_best_medians_df = df.sort_values(by='rsq_median', ascending=False)
x_df = sorted_rsq_best_medians_df[sorted_rsq_best_medians_df.rsq_median > r_sqr_median_cutoff]
print('{} SNPs with r_sqr > {:5.3f}'.format(x_df.shape[0], r_sqr_median_cutoff))
taxon_table_df = read_taxon_file(taxon_table_file_path, transform=transform)
for row_i in range(sorted_rsq_best_medians_df.shape[0]):
if row_i >= snp_count:
break
else:
# get a 1-row dataframe
snp_df = sorted_rsq_best_medians_df.iloc[[row_i]]
aligned_snp_df, aligned_taxa_df = align_snp_and_taxa(
snp_df,
taxon_table_df
)
# get the taxon stability selection scores
# use the taxon table df index to get column names for snp_df
taxon_scores_df = snp_df.loc[:, taxon_table_df.index].transpose()
sorted_taxon_scores_df = taxon_scores_df.sort_values(by=taxon_scores_df.columns[0], ascending=False)
#sorted_taxon_scores_df = taxon_scores_df.sort(taxon_scores_df.columns[0], ascending=False)
p_df_list = []
print('{} {} {:5.3f}'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID, snp_df.iloc[0].rsq_median))
summary_line = '{}\t{}\t'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID)
for i, (selected_taxon, selected_taxon_row) in enumerate(sorted_taxon_scores_df.iterrows()):
# use selected_taxon_row.index[0] to index the first and only column
selected_taxon_score = selected_taxon_row.iloc[0]
if selected_taxon_score < stability_cutoff:
#print('done with selected taxa')
break
else:
# trim 'Root;' from the front of the taxon name
if selected_taxon.startswith('Root;'):
taxon_name = selected_taxon[5:]
else:
taxon_name = selected_taxon
print(' {:5.3f} {}'.format(selected_taxon_score, taxon_name))
summary_line += '{}, '.format(taxon_name)
gts = [
snp_df.iloc[0].REF + snp_df.iloc[0].REF, # 0
snp_df.iloc[0].REF + snp_df.iloc[0].ALT, # 1
snp_df.iloc[0].ALT + snp_df.iloc[0].ALT # 2
]
aligned_snp_value_list = aligned_snp_df.values.flatten().tolist()
data_dict = {
'chromosome': [snp_df.iloc[0].CHROM] * aligned_snp_df.shape[1],
'snp_id': [snp_df.iloc[0].ID] * aligned_snp_df.shape[1],
'gene': [snp_df.iloc[0].GENE] * aligned_snp_df.shape[1],
'taxon': [selected_taxon] * aligned_snp_df.shape[1],
'abundance': aligned_taxa_df[selected_taxon].values.tolist(),
'variant_allele_count': [str(int(v)) for v in aligned_snp_value_list],
'genotype': [gts[int(v)] for v in aligned_snp_value_list],
'sample_id' : aligned_snp_df.columns
}
columns_to_display = ['abundance', 'variant_allele_count', 'genotype', 'sample_id']
if extra_columns:
for extra_column in extra_columns.split(','):
data_dict[extra_column] = snp_df.iloc[0][extra_column]
columns_to_display.append(extra_column)
p_df = pd.DataFrame(data_dict)
p_df_list.append(p_df)
if no_tables:
pass
else:
p_df[columns_to_display].to_csv(
sys.stdout,
sep='\t'
)
# save a stacked bar plot
if len(p_df_list) > 0:
file_name = 'stacked_bar_plot_selected_taxa_{}_{}.pdf'.format(
snp_df.iloc[0].GENE,
snp_df.iloc[0].ID
)
p_df = pd.concat(p_df_list, axis=0)
# at this point the index for p_df looks like
# 0...76.0...76.0...76
# replace the index
p_df.index = range(p_df.shape[0])
#p_df.to_csv(file_path, sep='\t')
stacked_bar_title = '{}\n{}'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID)
def box_bar_lasso_lars_cv_C_stability_selection_features(
rvcf_input_file_path, taxon_table_file_path, transform, plot_output_dir_path, stability_cutoff, snp_count):
print('plotting {} SNPs from {}'.format(snp_count, rvcf_input_file_path))
if os.path.exists(plot_output_dir_path):
# delete it
print('deleting old plots')
shutil.rmtree(plot_output_dir_path)
os.makedirs(plot_output_dir_path)
# read the rvcf file and sort by rsq_median
df = pd.read_csv(rvcf_input_file_path, sep='\t')
sorted_rsq_best_medians_df = df.sort_values(by='rsq_median', ascending=False)
taxon_table_df = read_taxon_file(taxon_table_file_path, transform=transform)
# these are proxies for R functions
taxon_abundance_box_plot = get_taxon_abundance_box_plot()
taxon_abundance_stacked_bar_plot = get_taxon_abundance_stacked_bar_plot()
for row_i in range(sorted_rsq_best_medians_df.shape[0]):
if row_i >= snp_count:
break
else:
# get a 1-row dataframe
snp_df = sorted_rsq_best_medians_df.iloc[[row_i]]
aligned_snp_df, aligned_taxa_df = align_snp_and_taxa(
snp_df,
taxon_table_df
)
# get the taxon stability selection scores
# use the taxon table df index to get column names for snp_df
taxon_scores_df = snp_df.loc[:, taxon_table_df.index].transpose()
sorted_taxon_scores_df = taxon_scores_df.sort(taxon_scores_df.columns[0], ascending=False)
# print all sorted taxon scores to verify they are sorted high to low
p_df_list = []
summary_line = '{}\t{}\t'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID)
for i, (selected_taxon, selected_taxon_row) in enumerate(sorted_taxon_scores_df.iterrows()):
# use selected_taxon_row.index[0] to index the first and only column
selected_taxon_score = selected_taxon_row.iloc[0]
if selected_taxon_score < stability_cutoff:
break
else:
# trim 'Root;' from the front of the taxon name
if selected_taxon.startswith('Root;'):
taxon_name = selected_taxon[5:]
else:
taxon_name = selected_taxon
summary_line += '{}, '.format(taxon_name)
# print a box plot
r_pdf_file_path = \
os.path.join(
plot_output_dir_path,
'best_taxa_{}_{}_{}_boxplot_{}.pdf'.format(
row_i,
snp_df.iloc[0].GENE,
snp_df.iloc[0].ID,
i
)
)
#print('writing file {}'.format(r_pdf_file_path))
gts = [
snp_df.iloc[0].REF + snp_df.iloc[0].REF, # 0
snp_df.iloc[0].REF + snp_df.iloc[0].ALT, # 1
snp_df.iloc[0].ALT + snp_df.iloc[0].ALT # 2
]
aligned_snp_value_list = aligned_snp_df.values.flatten().tolist()
p_df = pd.DataFrame({
'chromosome': [snp_df.iloc[0].CHROM] * aligned_snp_df.shape[1],
'snp_id': [snp_df.iloc[0].ID] * aligned_snp_df.shape[1],
'gene': [snp_df.iloc[0].GENE] * aligned_snp_df.shape[1],
'taxon': [selected_taxon] * aligned_snp_df.shape[1],
'abundance': aligned_taxa_df[selected_taxon].values.tolist(),
'variant_allele_count': [str(int(v)) for v in aligned_snp_value_list],
'gt': [gts[int(v)] for v in aligned_snp_value_list]
})
p_df_list.append(p_df)
r_df = rpy2.robjects.vectors.DataFrame({
'abundance': rpy2.robjects.FloatVector(aligned_taxa_df[selected_taxon].values.tolist()),
'variant_allele_count': rpy2.robjects.StrVector([str(int(v)) for v in aligned_snp_value_list]),
'genotype': rpy2.robjects.StrVector([gts[int(v)] for v in aligned_snp_value_list])
})
print(taxon_name)
print(r_df)
taxon_abundance_box_plot(
r_df,
r_pdf_file_path,
'{} (score: {:4.3f})'.format(snp_df.iloc[0].GENE, selected_taxon_score),
'{} {}'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID),
selected_taxon
)
# write a summary line and
print(summary_line[:-2])
#summary_file.write(summary_line[:-2])
#summary_file.write('\n')
# save a stacked bar plot
if len(p_df_list) > 0:
file_name = 'stacked_bar_plot_selected_taxa_{}_{}.pdf'.format(
snp_df.iloc[0].GENE,
snp_df.iloc[0].ID
)
stacked_bar_plot_file_path = os.path.join(plot_output_dir_path, file_name)
p_df = pd.concat(p_df_list, axis=0)
# at this point the index for p_df looks like
# 0...76.0...76.0...76
# replace the index
p_df.index = range(p_df.shape[0])
r_all_df = rpy2.robjects.vectors.DataFrame({
'abundance': rpy2.robjects.FloatVector(p_df['abundance'].values.tolist()),
'variant_allele_count': rpy2.robjects.StrVector([str(int(v)) for v in p_df['variant_allele_count'].values]),
'taxon': rpy2.robjects.StrVector(p_df['taxon']),
'gene': rpy2.robjects.StrVector(p_df['gene']),
'genotype': rpy2.robjects.StrVector(p_df['gt'])
})
stacked_bar_title = '{}\n{}'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID)
taxon_abundance_stacked_bar_plot(
r_all_df,
stacked_bar_plot_file_path,
stacked_bar_title,
'{} {}'.format(snp_df.iloc[0].GENE, snp_df.iloc[0].ID),
'median abundance'
)