def OutputSHMsPerPosition(self, output_fname):
aa_len = len(self.aa_dict.GetAAByIndex(self.GetRootIndex()))
num_shms = [0] * aa_len
for shm in self.shms.SHMIter():
num_shms[shm.pos] += self.shms.GetSHMMultiplicity(
shm) * self.shms.GetSHMMultiplicity(shm)
plt.figure()
ax = plt.gca()
ax.add_patch(
Rectangle((self.shms.cdr1_bounds[0], 0),
self.shms.cdr1_bounds[1] - self.shms.cdr1_bounds[0],
max(num_shms),
facecolor='#FFB4B6'))
ax.add_patch(
Rectangle((self.shms.cdr2_bounds[0], 0),
self.shms.cdr2_bounds[1] - self.shms.cdr2_bounds[0],
max(num_shms),
facecolor='#FFB4B6'))
ax.add_patch(
Rectangle((self.shms.cdr3_bounds[0], 0),
self.shms.cdr3_bounds[1] - self.shms.cdr3_bounds[0],
max(num_shms),
facecolor='#FFB4B6'))
plt.bar(range(aa_len), num_shms)
plt.xlabel('AA position')
plt.ylabel('Sum of squared SHM multiplicities')
plt.savefig(output_fname + '.svg')
utils.OutputPlotToPdf(output_fname + '.pdf')
def ComputeVariabilityPlot(self, output_fname):
# for each aa position compute how many aa in it
self.variability_sets = []
for i in range(self.most_freq_len):
self.variability_sets.append(set())
num_used_seq = 0
num_good_seq = 0
for aa in self.aa_dict:
if len(aa) != self.most_freq_len:
continue
num_good_seq += 1
for i in range(self.most_freq_len):
self.variability_sets[i].add(aa[i])
num_used_seq += len(self.aa_dict[aa])
self.variability = [
len(self.variability_sets[i]) for i in range(self.most_freq_len)
]
fig, ax = plt.subplots(1)
self._AddCDRsOnAminoAcidPlot(ax, max(self.variability) + 0.25)
plt.bar(range(self.most_freq_len), self.variability)
plt.ylim(1, max(self.variability) + 0.25)
plt.xlabel('Amino-acid position')
plt.ylabel('# amino-acids')
plt.title(str(num_used_seq) + ' aa seq were used')
utils.OutputPlotToPdf(output_fname)
def _OutputSHMsForV(self, v_gene, output_fname):
max_length = 300
num_roots = len(self.gene_usage[dataset.AnnotatedGene.V][v_gene])
if num_roots < 5:
return
for shms in self.gene_usage[dataset.AnnotatedGene.V][v_gene]:
for shm in shms:
max_length = max(max_length, shm.pos)
pos_mult = [0] * max_length
for shms in self.gene_usage[dataset.AnnotatedGene.V][v_gene]:
for shm in shms:
if shm.IsSubstitution():
pos_mult[shm.pos] += 1
plt.bar(range(max_length), pos_mult)
plt.title(v_gene + ', ' + str(num_roots) + ' roots')
plt.xlabel('Position (nt)')
plt.ylabel('# roots')
utils.OutputPlotToPdf(output_fname)
def OutputGraphSHMsAsMatrix(self, vertex_orderer, output_base):
if self.NumVertices() < 2:
return
matrix = []
#vertex_order = vertex_orderer.GetOrder()
aa_len = len(self.aa_dict.GetAAByIndex(self.GetRootIndex()))
for e in self.EdgeIter():
matrix.append([0] * aa_len)
for i in range(len(matrix)):
for j in range(len(matrix[i])):
if self._PositionIsInCDRs(j):
matrix[i][j] = 1
edge_ind = 0
for e in self.EdgeIter():
src_aa = self.aa_dict.GetAAByIndex(e[0])
dst_aa = self.aa_dict.GetAAByIndex(e[1])
for i in range(len(src_aa)):
if src_aa[i] != dst_aa[i]:
matrix[edge_ind][i] = 2
edge_ind += 1
# for i in range(1, len(vertex_order)):
# cur_aa = self.aa_dict.GetAAByIndex(vertex_order[i])
# parent_aa = self.aa_dict.GetAAByIndex(self.GetIncomingVertices(vertex_order[i])[0])
# for j in range(len(cur_aa)):
# if cur_aa[j] != parent_aa[j]:
# matrix[i][j] = 2
# vertex_levels = clonal_graph_algorithms.GetLevelsByVertexOrder(self, vertex_order)
# level_colors = []
# for l in vertex_levels:
# level_colors.append(utils.GetColorByNormalizedValue('prism', float(l) / max(vertex_levels)))
sns.heatmap(matrix,
cmap='coolwarm',
xticklabels=[],
yticklabels=[],
cbar=False)
# sns.clustermap(matrix, cmap = 'coolwarm', yticklabels = [str(v) for v in vertex_order], row_colors = level_colors, row_cluster = False, col_cluster = False, xticklabels = [])
plt.savefig(output_base + ".svg")
utils.OutputPlotToPdf(output_base + '.pdf')
def OutputNumCodonsPerAAPosition(self, output_fname):
tree = self.full_length_lineage.UndirectedClonalTree()
most_frequent_aa = [''] * self.most_freq_len
for i in range(self.most_freq_len):
cur_aa_dict = dict()
for v in tree.VertexIter():
v_aa_seq = self.aa_dict.GetAAById(
self.full_length_lineage.GetFullLengthSequenceByIndex(
v).id)
if len(v_aa_seq) != self.most_freq_len:
continue
if v_aa_seq[i] not in cur_aa_dict:
cur_aa_dict[v_aa_seq[i]] = 0
cur_aa_dict[v_aa_seq[i]] += 1
most_frequent_aa[i] = max(cur_aa_dict.iteritems(),
key=operator.itemgetter(1))[0]
codons = [0] * self.most_freq_len
for i in range(self.most_freq_len):
cur_codons = set()
for v in tree.VertexIter():
v_nucl_seq = self.full_length_lineage.GetFullLengthSequenceByIndex(
v).seq
v_aa_seq = self.aa_dict.GetAAById(
self.full_length_lineage.GetFullLengthSequenceByIndex(
v).id)
if len(v_aa_seq) != self.most_freq_len:
continue
if v_aa_seq[i] != most_frequent_aa[i]:
continue
cur_codons.add(v_nucl_seq[i * 3:i * 3 + 3])
codons[i] = len(cur_codons)
#print codons
fig, ax = plt.subplots(1)
self._AddCDRsOnAminoAcidPlot(ax, max(codons) + 0.25)
plt.bar(range(self.most_freq_len), codons)
plt.ylim(1, max(codons) + 0.25)
utils.OutputPlotToPdf(output_fname)