def run_threshold_analysis(output_directory, title, input_repertoires, trim_first=True, rawtext=True, k=1):
all_seqs = []
# read in the clusters
repertoire_dict = {}
cluster_number_to_rep = {}
i = 0
for file in input_repertoires:
i += 1
cluster_number_to_rep[i] = file
content = txttoseqlist(file, trim_first=trim_first, rawtext=rawtext)
all_seqs += content
for cdr in content:
if cdr in repertoire_dict:
found = False
for nums in repertoire_dict[cdr]:
if nums == i:
found = True
if not found:
repertoire_dict[cdr].append(i)
else:
repertoire_dict[cdr] = [i]
# get cluster length
set_content = set(content)
num_clusters = i
print(generate_threshold_plot(all_seqs, repertoire_dict, k, num_clusters, output_directory, title, step=.05, cluster_number_to_rep=cluster_number_to_rep))
def twins_threshold_analysis(output_directory,
twinA1,
twinA2,
twinC1,
twinC2,
twinD1,
twinD2,
beta=True):
twin_pair_A = txttoseqlist(twinA1, trim_first=True) + txttoseqlist(
twinA2, trim_first=True)
twin_pair_C = txttoseqlist(twinC1, trim_first=True) + txttoseqlist(
twinC2, trim_first=True)
twin_pair_D = txttoseqlist(twinD1, trim_first=True) + txttoseqlist(
twinD2, trim_first=True)
# run threshold analysis on each twin pair
if beta:
run_threshold_analysis_repertoire_lists(
output_directory,
"TwA_Beta", [twin_pair_A, twin_pair_C + twin_pair_D],
k=1)
run_threshold_analysis_repertoire_lists(
output_directory,
"TwC_Beta", [twin_pair_C, twin_pair_A + twin_pair_D],
k=1)
run_threshold_analysis_repertoire_lists(
output_directory,
"TwD_Beta", [twin_pair_D, twin_pair_A + twin_pair_C],
k=1)
else:
run_threshold_analysis_repertoire_lists(
output_directory,
"TwA_Alpha", [twin_pair_A, twin_pair_C + twin_pair_D],
k=1)
run_threshold_analysis_repertoire_lists(
output_directory,
"TwC_Alpha", [twin_pair_C, twin_pair_A + twin_pair_D],
k=1)
run_threshold_analysis_repertoire_lists(
output_directory,
"TwD_Alpha", [twin_pair_D, twin_pair_A + twin_pair_C],
k=1)
fig.savefig(output_directory + title + "_kde_updated.pdf")
plt.figure() # new figure
figure = sns.kdeplot(self_scores, shade=True, cumulative=True)
figure = sns.kdeplot(non_self_scores, shade=True, cumulative=True)
figure.set(ylim=(0, 1))
fig = figure.get_figure()
fig.savefig(output_directory + title + "_kde_cumulative_updated.pdf")
print("calculating self-scores...")
# calculate self-scores
all_self_scores = []
num_self_scores = 0
for file in cluster_file_list:
cdrs = txttoseqlist(file)
cdrs = set(cdrs)
cdrs = list(cdrs)
scores = calculate_min_distances_within(cdrs, cdrs)
all_self_scores += scores
num_self_scores += len(scores)
print(all_self_scores)
print("calculating non-self scores...")
# calculate non-self scores
all_non_self_scores = []
num_non_self_scores = 0
for file1 in cluster_file_list:
for file2 in cluster_file_list:
if file1 != file2:
cdrs1 = list(set(txttoseqlist(file1)))
"need the following arguments: output directory, threshold to cluster to, "
"human/mice, which epitope to get the motifs of, data directory")
print("human epitopes: BMLF1, p65, M1")
print("mice epitopes: PB1, PB1-F2, NP, PA, M38, M45, m139")
sys.exit()
output_dir = sys.argv[1]
distance = float(sys.argv[2])
human = str(sys.argv[3])
epitope = sys.argv[4]
data_dir = sys.argv[5] # datasets/dash_data/
epitope_filename = data_dir + epitope + "_beta.txt"
print(epitope_filename)
Xlabels = txttoseqlist(epitope_filename, trim_first=True, remove_special=True)
human_files = [
data_dir + "BMLF1_beta.txt", data_dir + "p65_beta.txt",
data_dir + "M1_beta.txt"
]
mice_files = [
data_dir + "PB1_beta.txt", data_dir + "PB1-F2_beta.txt",
data_dir + "NP_beta.txt", data_dir + "PA_beta.txt",
data_dir + "M38_beta.txt", data_dir + "M45_beta.txt",
data_dir + "m139_beta.txt"
]
other_files = []
if human.lower() == 'true':
for file in human_files:
sys.exit()
else:
output_directory = sys.argv[1]
cdr_type = sys.argv[2]
repertoire_file_list = sys.argv[3].split(',')
all_seqs = []
# read in the clusters
seq_to_rep = {}
rep_to_seq = {}
i = 0
for file in repertoire_file_list:
i += 1
content = txttoseqlist(file, trim_first=False)
print(file)
print(len(content))
rep_to_seq[i] = content
all_seqs += content
for cdr in content:
if cdr in seq_to_rep:
found = False
for nums in seq_to_rep[cdr]:
if nums == i:
found = True
if not found:
seq_to_rep[cdr].append(i)
else:
seq_to_rep[cdr] = [i]
def cluster_and_get_distance_scores(subject_file, subject_twin_file,
other_files, clustering_distance,
output_dir):
log_file = open(output_dir + "output.txt", mode='w')
log_file.write(subject_file[11:])
Xlabels = txttoseqlist(subject_file, trim_first=True, remove_special=True)
Xlabelstwin = txttoseqlist(subject_twin_file,
trim_first=True,
remove_special=True)
print("clustering subject and their twin...")
clusters = cluster_data(Xlabels, max_distance=clustering_distance)
clusters_twin = cluster_data(Xlabelstwin, max_distance=clustering_distance)
print("clustered subject and their twin")
# convert clusters to list of clusters
clusters_list = []
for key in clusters:
clusters_list.append(clusters[key])
twin_clusters_list = []
for key in clusters_twin:
twin_clusters_list.append(clusters_twin[key])
print("clustering other individuals...")
other_clusterings = []
# get the other twins reperotires
for file in other_files:
Xlabelsother = txttoseqlist(file, trim_first=True, remove_special=True)
clusters2other = cluster_data(Xlabelsother,
max_distance=clustering_distance)
other_clusters = []
for key in clusters2other:
other_clusters.append(clusters2other[key])
other_clusterings.append(other_clusters)
print("clustered others")
print("getting scores...")
# get the scores
scores = score_repertoire_twins(clusters_list,
twin_clusters_list,
other_clusterings,
min_clust_size=4)
print(scores)
n = 1
for score_data in scores:
log_file.write("\n\ncluster #" + str(n))
# print(score_data)
# print("parsing data...\n")
# parse score data
cluster = score_data[1]
twin_cluster = score_data[0][1]
twin_cluster_score = score_data[0][0]
log_file.write("\ncluster is :" + str(cluster))
log_file.write("\nthe closest twin cluster is " + str(twin_cluster))
log_file.write("\ntheir score is: " + str(twin_cluster_score))
# create motifs
create_motif(cluster, str(n) + "_subject", output_dir)
create_motif(twin_cluster, str(n) + "_subject_twin", output_dir)
# calculate average distance to other clusters
total_dist = 0
count = 0
other_scores = []
for i in range(len(other_files)):
data = score_data[i + 2]
repertoire = other_files[i][22:]
other_score = data[0]
other_cluster = data[2]
log_file.write("\nscore with " + str(repertoire) + " is " +
str(other_score))
log_file.write("\ncluster: " + str(other_cluster))
create_motif(other_cluster,
str(n) + "_" + str(repertoire), output_dir)
total_dist += other_score
count += 1
other_scores.append(other_score)
avg_score = total_dist / count
log_file.write("\naverage score to another cluster is " +
str(avg_score))
log_file.write("\nthe closest score to another cluster is " +
str(min(other_scores)))
log_file.write(' ')
n += 1