for ni,node in enumerate(prediction.non_terminals):
node.name = str(ni+1)
# write tree to file
Phylo.write(prediction.T, dirname+'/reconstructed_tree.nwk', 'newick')
# write inferred ancestral sequences to file
with open(dirname+'/ancestral_sequences.fasta', 'w') as outfile:
for node in prediction.non_terminals:
outfile.write('>'+node.name+'\n'+str(node.seq)+'\n')
## write sequence ranking to file
# terminal nodes
prediction.rank_by_method(nodes = prediction.terminals, method = 'mean_fitness');
with open(dirname+'/sequence_ranking_terminals.txt', 'w') as outfile:
outfile.write('#'+'\t'.join(['name','rank', 'mean', 'standard dev'])+'\n')
for node in prediction.terminals:
outfile.write('\t'.join(map(str,[node.name, node.rank, node.mean_fitness, np.sqrt(node.var_fitness)]))+'\n')
# terminal nodes
prediction.rank_by_method(nodes = prediction.non_terminals, method = 'mean_fitness');
with open(dirname+'/sequence_ranking_nonterminals.txt', 'w') as outfile:
outfile.write('#'+'\t'.join(['name','rank', 'mean', 'variance'])+'\n')
for node in prediction.non_terminals:
outfile.write('\t'.join(map(str,[node.name, node.rank, node.mean_fitness, np.sqrt(node.var_fitness)]))+'\n')
if params.plot:
tree_utils.plot_prediction_tree(prediction)
plt.savefig(dirname+'/marked_up_tree.pdf')
Phylo.write(prediction.T, dirname + '/reconstructed_tree.nwk', 'newick')
# write inferred ancestral sequences to file
with open(dirname + '/ancestral_sequences.fasta', 'w') as outfile:
for node in prediction.non_terminals:
outfile.write('>' + node.name + '\n' + str(node.seq) + '\n')
## write sequence ranking to file
# terminal nodes
prediction.rank_by_method(nodes=prediction.terminals, method='polarizer')
with open(dirname + '/sequence_ranking_terminals.txt', 'w') as outfile:
outfile.write('#' + '\t'.join(['name', 'rank', 'LBI']) + '\n')
for node in prediction.terminals:
outfile.write(
'\t'.join(map(str, [node.name, node.rank, node.polarizer])) + '\n')
# terminal nodes
prediction.rank_by_method(nodes=prediction.non_terminals, method='polarizer')
with open(dirname + '/sequence_ranking_nonterminals.txt', 'w') as outfile:
outfile.write('#' + '\t'.join(['name', 'rank', 'LBI']) + '\n')
for node in prediction.non_terminals:
outfile.write(
'\t'.join(map(str, [node.name, node.rank, node.polarizer])) + '\n')
# plot the tree if desired
if params.plot:
tree_utils.plot_prediction_tree(prediction,
method='polarizer',
internal=True)
plt.savefig(dirname + '/marked_up_tree.pdf')
node.name = str(ni+1)
# write tree to file
Phylo.write(prediction.T, dirname+'/reconstructed_tree.nwk', 'newick')
# write inferred ancestral sequences to file
with open(dirname+'/ancestral_sequences.fasta', 'w') as outfile:
for node in prediction.non_terminals:
outfile.write('>'+node.name+'\n'+str(node.seq)+'\n')
## write sequence ranking to file
# terminal nodes
prediction.rank_by_method(nodes = prediction.terminals, method = 'polarizer');
with open(dirname+'/sequence_ranking_terminals.txt', 'w') as outfile:
outfile.write('#'+'\t'.join(['name','rank', 'LBI'])+'\n')
for node in prediction.terminals:
outfile.write('\t'.join(map(str,[node.name, node.rank, node.polarizer]))+'\n')
# terminal nodes
prediction.rank_by_method(nodes = prediction.non_terminals, method = 'polarizer');
with open(dirname+'/sequence_ranking_nonterminals.txt', 'w') as outfile:
outfile.write('#'+'\t'.join(['name','rank', 'LBI'])+'\n')
for node in prediction.non_terminals:
outfile.write('\t'.join(map(str,[node.name, node.rank, node.polarizer]))+'\n')
# plot the tree if desired
if params.plot:
tree_utils.plot_prediction_tree(prediction, method='polarizer', internal=True)
plt.savefig(dirname+'/marked_up_tree.pdf')
method = methods[0]
color_internal = True
if params.flutype.startswith('H3N2') and year in laessig_prediction:
seq_labels = {
prediction.best_node(method=method[0], nodes=method[2]).name: '*',
laessig_prediction[year].name: "L&L"
}
else:
seq_labels = {prediction.best_node().name: '*'}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.erase_color(prediction.T)
tree_utils.erase_color(combined_data.T)
tree_utils.plot_prediction_tree(prediction, internal=color_internal)
plt.title(
"predicting " + params.flutype + " season " + str(year) + "/" +
str(year + 1) + ": " + str(
np.round(distances[method[0] + method[1]] /
distances['average'], 4)))
#plt.savefig('../figures/'+base_name+'_prediction_'+method[0]+method[1]+'_'+name_mod+'.pdf')
# plot a combined figure
# color according to season
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
for c in combined_data.T.get_terminals():
if c.name in pred_names:
c.color = (178, 34, 34)
else:
otherseqsnames=[]
combined_data = test_flu.make_combined_data(prediction, test_data, otherseqsnames)
seq_labels = {prediction.best_node().name:'*'} #, laessig_prediction[year].name:"L&L"}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.label_nodes(combined_data.T, seq_labels)
# plot a combined figure
fig = plt.figure(figsize = (12,6))
#subplot 1: only the prediction data
ax = plt.subplot(121)
#add panel label
plt.text(-0.06,0.95,'A', transform = plt.gca().transAxes, fontsize = 36)
plt.title('until Feb '+str(year))
plt.tight_layout()
tree_utils.plot_prediction_tree(prediction, axes=ax, cb=True, offset = 0.0005, internal=False)
#subplot 2: prediction data and test data
ax = plt.subplot(122)
#add panel label
plt.text(-0.06,0.95,'B', transform = plt.gca().transAxes, fontsize = 36)
plt.title('until Feb '+str(year)+' + season '+str(year)+"/"+str(year+1)+" (grey)")
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
prediction.color_other_tree(combined_data.T.get_terminals(), offset = 0.0005)
for c in combined_data.T.get_terminals():
if c.name in pred_names:
pass
# c.color = (178, 34, 34 )
else:
c.color = (0,255,255)
# set up the filtering criteria and select sequences from the master alignment
criteria = [(date(year - 1, 5, 1), date(year, 2,
28), [region], sample_size)
for region in prediction_regions]
my_flu_alignment = flu_alignment(aln_fname,
outgroup,
annotation,
criteria=criteria,
cds={
'begin': 0,
'end': 987,
'pad': 0
})
# run the prediction
prediction = flu_ranking(my_flu_alignment, boost=0.5)
top_seq = prediction.predict()
print top_seq
# plot the tree colored by the prediction
tree_utils.plot_prediction_tree(prediction)
# plot the distribution of sampling dates
y, x, bin_names = my_flu_alignment.sampling_distribution(bins=10)
import matplotlib.pyplot as plt
plt.figure()
plt.plot(x[1:], y)
plt.xticks(x[1:], map(str, [b for b in bin_names]), rotation=30)
prediction.best_node('polarizer').name: '*'
} #, laessig_prediction[year].name:"L&L"}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.label_nodes(combined_data.T, seq_labels)
# plot a combined figure
fig = plt.figure(figsize=(12, 6))
#subplot 1: only the prediction data
ax = plt.subplot(121)
#add panel label
plt.text(-0.06, 0.95, 'A', transform=plt.gca().transAxes, fontsize=36)
plt.title('until Feb ' + str(year))
plt.tight_layout()
tree_utils.plot_prediction_tree(prediction,
method='polarizer',
axes=ax,
cb=True,
offset=0.0005,
internal=False)
#subplot 2: prediction data and test data
ax = plt.subplot(122)
#add panel label
plt.text(-0.06, 0.95, 'B', transform=plt.gca().transAxes, fontsize=36)
plt.title('until Feb ' + str(year) + ' + season ' + str(year) + "/" +
str(year + 1) + " (grey)")
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
prediction.color_other_tree(combined_data.T.get_terminals(),
method='polarizer',
offset=0.0005)
for c in combined_data.T.get_terminals():
prediction = sequence_ranking.sequence_ranking(
seq_data,
eps_branch_length=eps_branch_length,
pseudo_count=5,
methods=['mean_fitness'],
D=diffusion,
distance_scale=distance_scale,
samp_frac=samp_frac)
best_node = prediction.predict()
# Write fitness tree to file (for mutation annotation)
Phylo.write(prediction.T, outfile_fitness_tree, "newick")
# Get fitness changes on each branch
df_fitness = get_fitness_changes(prediction)
df_fitness.sort_values("delta_mean_fitness", ascending=False, inplace=True)
# Write fitness changes to file
df_fitness.to_csv(outfile_df_fitness)
# Plot tree colored by fitness with and without node labels
tree_utils.plot_prediction_tree(prediction)
plt.savefig(outfile_tree_pdf)
tree_utils.plot_prediction_tree(prediction,
node_label_func=lambda x: x.name)
plt.savefig(outfile_tree_pdf_labeled)
print "Done!!"
prediction.best_node().name: '*'
} #, laessig_prediction[year].name:"L&L"}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.label_nodes(combined_data.T, seq_labels)
# plot a combined figure
fig = plt.figure(figsize=(12, 6))
#subplot 1: only the prediction data
ax = plt.subplot(121)
#add panel label
plt.text(-0.06, 0.95, 'A', transform=plt.gca().transAxes, fontsize=36)
plt.title('until Feb ' + str(year))
plt.tight_layout()
tree_utils.plot_prediction_tree(prediction,
axes=ax,
cb=True,
offset=0.0005,
internal=False)
#subplot 2: prediction data and test data
ax = plt.subplot(122)
#add panel label
plt.text(-0.06, 0.95, 'B', transform=plt.gca().transAxes, fontsize=36)
plt.title('until Feb ' + str(year) + ' + season ' + str(year) + "/" +
str(year + 1) + " (grey)")
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
prediction.color_other_tree(combined_data.T.get_terminals(), offset=0.0005)
for c in combined_data.T.get_terminals():
if c.name in pred_names:
pass
# combine the test data, the prediction data and possible other sequences
# and build a tree
combined_data = test_flu.make_combined_data(prediction, test_data, otherseqsnames)
method = methods[0]
color_internal = True
if params.flutype.startswith('H3N2') and year in laessig_prediction:
seq_labels = {prediction.best_node(method= method[0], nodes = method[2]).name:'*', laessig_prediction[year].name:"L&L"}
else:
seq_labels = {prediction.best_node().name:'*'}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.erase_color(prediction.T)
tree_utils.erase_color(combined_data.T)
tree_utils.plot_prediction_tree(prediction, internal=color_internal)
plt.title("predicting "+params.flutype+" season "+str(year)+"/"+str(year+1)+": "+
str(np.round(distances[method[0]+method[1]]/distances['average'],4)))
#plt.savefig('../figures/'+base_name+'_prediction_'+method[0]+method[1]+'_'+name_mod+'.pdf')
# plot a combined figure
# color according to season
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
for c in combined_data.T.get_terminals():
if c.name in pred_names:
c.color = (178, 34, 34 )
else:
c.color = (0,255,255)
prediction.interpolate_color(combined_data.T)
otherseqsnames=[]
combined_data = test_flu.make_combined_data(prediction, test_data, otherseqsnames)
seq_labels = {prediction.best_node('polarizer').name:'*'} #, laessig_prediction[year].name:"L&L"}
tree_utils.label_nodes(prediction.T, seq_labels)
tree_utils.label_nodes(combined_data.T, seq_labels)
# plot a combined figure
fig = plt.figure(figsize = (12,6))
#subplot 1: only the prediction data
ax = plt.subplot(121)
#add panel label
plt.text(-0.06,0.95,'A', transform = plt.gca().transAxes, fontsize = 36)
plt.title('until Feb '+str(year))
plt.tight_layout()
tree_utils.plot_prediction_tree(prediction, method='polarizer', axes=ax, cb=True, offset = 0.0005, internal=False)
#subplot 2: prediction data and test data
ax = plt.subplot(122)
#add panel label
plt.text(-0.06,0.95,'B', transform = plt.gca().transAxes, fontsize = 36)
plt.title('until Feb '+str(year)+' + season '+str(year)+"/"+str(year+1)+" (grey)")
pred_names = [c.name for c in prediction.terminals]
tree_utils.erase_color(combined_data.T)
prediction.color_other_tree(combined_data.T.get_terminals(), method='polarizer', offset = 0.0005)
for c in combined_data.T.get_terminals():
if c.name in pred_names:
pass
# c.color = (178, 34, 34 )
else:
c.color = (0,255,255)
