def val_df_to_sharedness(value_df, headers):
"""
NB only for use with the phospho data
:param value_df: A pandas dataframe with samples for columns, peps/prots for rows, with vals = affinity/abundance
:param headers: headers for final table
:return: a longform dataframe for plotting
"""
vals = []
# Just in cultured
for thing in [
x for x in list(value_df.index)
if value_df.loc[x][0] > 0 and sum(value_df.loc[x][1:]) == 0
]:
vals.append(['C-only', thing] +
[x for x in value_df.loc[thing] if x > 0])
# Just in one single mouse
for thing in [
x for x in list(value_df.index)
if len([y for y in list(value_df.loc[x])
if y > 0]) == 1 and value_df.loc[x][0] == 0
]:
vals.append(['1m-only', thing] +
[x for x in value_df.loc[thing] if x > 0])
# In cultured and X #s mice mouse only
for i in range(2, 6):
for thing in [
x for x in list(value_df.index)
if len([y for y in list(value_df.loc[x])
if y > 0]) == i and value_df.loc[x][0] > 0
]:
for val in [x for x in value_df.loc[thing] if x > 0]:
vals.append(['C+' + str(i - 1) + 'm', thing, val])
return fxn.list_to_df(vals, headers, False)
#
properties = GlobalDescriptor(peptides[gp])
properties.charge_density(ph=7.4, amide=True)
charge_densities[gp] = [x[0] for x in properties.descriptor]
for val in charge_densities[gp]:
charge_densities_long.append([gp, val])
#
polarities[gp] = get_peptide_values(peptides[gp], 'polarity')
for val in polarities[gp]:
polarities_long.append([gp, val])
#
gravy[gp] = get_peptide_values(peptides[gp], 'gravy')
for val in gravy[gp]:
gravy_long.append([gp, val])
eisenbergs_long = fxn.list_to_df(eisenbergs_long, ['Group', 'Hydrophobicity'],
False)
charges_long = fxn.list_to_df(charges_long, ['Group', 'Charge'], False)
charge_densities_long = fxn.list_to_df(charge_densities_long,
['Group', 'Charge density'], False)
polarities_long = fxn.list_to_df(polarities_long, ['Group', 'Polarity'], False)
gravy_long = fxn.list_to_df(gravy_long, ['Group', 'Hydrophobicity'], False)
ordr = ['Cultured', 'Mouse', 'Both']
sns.violinplot(data=eisenbergs_long, x='Group', y='Hydrophobicity', order=ordr)
plt.xlabel("")
plt.ylabel("Hydrophobicity (Eisenberg scale)")
plt.savefig(plot_dir + 'eisenberg-hydrophobicity.png',
dpi=300,
bbox_inches='tight')
plt.close()
protein_dict[protein][nam] = prot_dat[nam][protein]
samples = pep_dat.keys()
samples.sort()
peptide_list = []
for peptide in peptide_dict:
# peptide_list.append([peptide, peptide_dict[peptide]['Hunt'], peptide_dict[peptide]['Cultured'],
peptide_list.append([
peptide, peptide_dict[peptide]['C'], peptide_dict[peptide]['M01'],
peptide_dict[peptide]['M04'], peptide_dict[peptide]['M07'],
peptide_dict[peptide]['M10']
])
peptides = fxn.list_to_df(peptide_list,
['Peptide', 'C', 'M01', 'M04', 'M07', 'M10'],
True)
protein_list = []
for protein in protein_dict:
protein_list.append([
protein, protein_dict[protein]['C'], protein_dict[protein]['M01'],
protein_dict[protein]['M04'], protein_dict[protein]['M07'],
protein_dict[protein]['M10']
])
proteins = fxn.list_to_df(protein_list,
['Protein', 'C', 'M01', 'M04', 'M07', 'M10'],
True)
# Plot total number of peptides detected
y_type = get_data_source(pep_samples[y])
comparison_type = [x_type, y_type]
comparison_type.sort()
if comparison_type != ['control', 'control']:
for jacc in sampled_jacc[x][y]:
sampled_jacc_long.append(
['-\n'.join(comparison_type), jacc])
whole_jacc_long.append([
'-\n'.join(comparison_type),
fxn.jaccard(peptides[pep_samples[x]],
peptides[pep_samples[y]])
])
done_comparisons.append(str(x) + '-' + str(y))
done_comparisons.append(str(y) + '-' + str(x))
sampled_jacc_long = fxn.list_to_df(sampled_jacc_long,
['Comparison', 'Jaccard'], False)
whole_jacc_long = fxn.list_to_df(whole_jacc_long,
['Comparison', 'Jaccard'], False)
col_order = [
'control-\npublished', 'control-\nculture', 'control-\nmouse',
'published-\npublished', 'culture-\npublished', 'mouse-\npublished',
'culture-\nmouse', 'mouse-\nmouse', 'culture-\nculture'
]
fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(111)
sns.violinplot(x='Comparison',
y='Jaccard',
data=sampled_jacc_long,
order=col_order,
# Read in data
weights = []
with open('../Data/sample-weights.csv', 'rU') as in_file:
line_count = 0
for line in in_file:
bits = line.rstrip().split(',')
if line_count == 0:
headers = bits
else:
weights.append([int(x) for x in bits[:2]] + [bits[2]])
line_count += 1
headers[0] = 'Day'
headers[2] = 'Growth type'
weights = fxn.list_to_df(weights, headers, False)
weights = weights.sort_values(by='Growth type')
fig = plt.figure(figsize=(5, 5))
ax = fig.add_subplot(111)
sns.scatterplot(data=weights, x='Day', y='mg', hue='Growth type')
plt.ylabel("Cumulative sample weight (mg)")
plt.xlabel("Days of growth")
plt.savefig(plot_dir + 'weight-over-time.png', dpi=300, bbox_inches='tight')
plt.close()
# Read in number of peptides to check correlation with input xenograft tumor weight
with open('../Data/mouse-weights-v-peptides.csv', 'rU') as in_file:
wvp = []
line_count = 0
# Go through the matches and count!
vals = []
positives = {}
totals = {}
col_order = ['Cultured', 'Mouse', 'Both']
for growth_type in col_order:
positive = len(
[x for x in matches[growth_type] if matches[growth_type][x] > 0])
total = len(matches[growth_type])
print growth_type, positive, total, positive / total
vals.append([growth_type, positive / total])
positives[growth_type] = positive
totals[growth_type] = total
vals = fxn.list_to_df(vals, ['Type', 'Proportion'], False)
# Finally plot
fig = plt.figure(figsize=(4.5, 5))
ax = fig.add_subplot(111)
sns.barplot(data=vals,
x='Type',
y='Proportion',
palette=fxn.c_m_mix_cols,
order=col_order)
for x in range(len(col_order)):
plot_text = str(positives[col_order[x]]) + '/\n' + str(
totals[col_order[x]])
print plot_text
ax.text(x,
.64,
elif best_allele == 'HLA-B0702':
allele_count['HLA-B*\n07:02'] += 1
elif best_allele == 'HLA-C0702':
allele_count['HLA-C*\n07:02'] += 1
else:
allele_count['NPB'] += 1 # No predicted binder
# Write out to nested list, to be turned into long proportions df, to be plotted as in peptide-analysis.py
type_key = {'C': 'Cultured', 'M': 'Mouse'}
for allele in allele_count:
predictions.append([
type_key[nam[0]], allele,
allele_count[allele] / sum(allele_count.values())
])
props = fxn.list_to_df(predictions, ['Type', 'Allele', 'Proportion'],
False)
hla_alleles = ['HLA-A*\n02:01', 'HLA-B*\n07:02', 'HLA-C*\n07:02', 'NPB']
# HLA allele contribution barplot
fig = plt.figure(figsize=(4.5, 5))
ax = fig.add_subplot(111)
sns.barplot(data=props,
x='Allele',
y='Proportion',
hue='Type',
order=hla_alleles)
sns.swarmplot(data=props,
x='Allele',
y='Proportion',
hue='Type',
len(peptides[cell_line][growth_type])
])
# Relative length distribution calculation
for peptide in peptides[cell_line][growth_type]:
lengths_raw[cell_line][growth_type][len(peptide)] += 1
for i in range(8, 16):
proportion = lengths_raw[cell_line][growth_type][i] / sum(
lengths_raw[cell_line][growth_type].values())
peptide_lens_props.append(
[cell_line, i, proportion, nam_key[growth_type]])
# Relative length distribution plotting
peptide_lens_props = fxn.list_to_df(
peptide_lens_props, ['Cell Line', 'Length', 'Proportion', 'Type'],
False)
fig = plt.figure(figsize=(3, 5))
ax = fig.add_subplot(111)
sns.barplot(x='Length',
y='Proportion',
hue='Type',
data=peptide_lens_props)
plt.xlabel("Length (amino acids)")
plt.savefig(plot_dir + cell_line + '-length-bars.png',
dpi=300,
bbox_inches='tight')
plt.close()
# Total overlap Venn (unsampled)