def find_clusters_and_gene_enrichment(tested_gene_list_file_name,
total_gene_list_file_name,
gene_expression_file_name,
phenotype_file_name,
gene_filter_file_name=None,
tested_gene_list_path=None,
total_gene_list_path=None,
gene_expression_path=None,
phenotype_path=None,
gene_filter_file_path=None,
var_th_index=None,
start_k=2,
end_k=6,
calc_go=True,
enrichment_list_file_names=None,
meta_groups=None,
filter_expression=None,
cluster_algorithm=None):
# fetch gene expression by gene_id, divided by tumor type
gene_sets = []
expression_sets = []
averaged_expression_sets = []
tested_gene_expression = load_gene_expression_profile_by_genes(
tested_gene_list_file_name, gene_expression_file_name,
gene_filter_file_name, tested_gene_list_path, gene_expression_path,
gene_filter_file_path)
tested_gene_expression_headers_rows, tested_gene_expression_headers_columns, tested_gene_expression = separate_headers(
tested_gene_expression)
if filter_expression is not None:
filtered_patients = [
y for x in divided_patient_ids_by_label(phenotype_file_name,
groups=filter_expression)
for y in x
]
print "number of filtered patients from phenotypes: {}".format(
len(filtered_patients))
else:
print "no filter applied"
filtered_patients = tested_gene_expression_headers_columns
tested_gene_expression, tested_gene_expression_headers_columns = filter_genes_dataset_by_patients(
filtered_patients, tested_gene_expression_headers_columns,
tested_gene_expression)
if np.shape(tested_gene_expression)[1] == 1:
print "no expressions were found after filtering by labels {}. skipping...".format(
filter_expression)
return None
total_gene_list = load_gene_list(total_gene_list_file_name)
tested_gene_list = load_gene_list(tested_gene_list_file_name)
row_var = np.var(tested_gene_expression, axis=1)
row_var_sorted = np.sort(row_var)[::-1]
labels_assignment_patients = None
if meta_groups is not None:
print "clustering patients by groups"
labels_assignment_patients = labels_assignments(
meta_groups, phenotype_file_name,
tested_gene_expression_headers_columns)
enrichment_lists = []
if enrichment_list_file_names is not None:
for cur in enrichment_list_file_names:
enrichment_lists.append(load_gene_list(cur))
if var_th_index is None:
var_th_index = len(row_var_sorted) - 1
row_var_th = row_var_sorted[var_th_index]
row_var_masked_indices = np.where(row_var_th > row_var)[0]
gene_expression_top_var = np.delete(tested_gene_expression,
row_var_masked_indices,
axis=0)
gene_expression_top_var_header_rows = np.delete(
tested_gene_expression_headers_rows, row_var_masked_indices, axis=0)
gene_expression_top_var_header_columns = tested_gene_expression_headers_columns
clfs_results = {}
output_rows = []
if calc_go:
if not os.path.exists(
os.path.join(constants.GO_DIR, constants.GO_FILE_NAME)):
wget.download(
constants.GO_OBO_URL,
os.path.join(constants.GO_DIR, constants.GO_FILE_NAME))
# if not os.path.exists(os.path.join(constants.TCGA_DATA_DIR, 'goa_human.gaf')):
# wget.download(go_obo_url, os.path.join(constants.TCGA_DATA_DIR, 'goa_human.gaf'))
obo_dag = GODag(os.path.join(constants.GO_DIR, constants.GO_FILE_NAME))
assoc = read_ncbi_gene2go(os.path.join(
constants.GO_DIR, constants.GO_ASSOCIATION_FILE_NAME),
no_top=True)
g = GOEnrichmentStudy(
[int(cur) for cur in ensembl2entrez_convertor(total_gene_list)],
assoc,
obo_dag,
methods=["bonferroni", "fdr_bh"])
g_res = g.run_study([
int(cur) for cur in ensembl2entrez_convertor(
gene_expression_top_var_header_rows)
])
GO_results = [(cur.NS, cur.GO, cur.goterm.name, cur.p_uncorrected,
cur.p_fdr_bh) for cur in g_res if cur.p_fdr_bh <= 0.05]
print GO_results
if cluster_algorithm == "kmeans":
for n_clusters in range(start_k, end_k + 1):
clfs_results[n_clusters] = []
centres, km_clf, dist = kmeanssample(X=gene_expression_top_var,
k=n_clusters,
metric="euclidean")
for i in range(n_clusters):
ranks = []
for j in range(n_clusters):
ranks.append(
np.average(
np.delete(gene_expression_top_var,
np.where(km_clf != j)[0],
axis=0)))
ranks = rankdata(ranks)
cluster_labels = np.array(km_clf)
for j in range(n_clusters):
cluster_labels[np.where(km_clf == ranks[j] - 1)] = j
labels_assignment = [cluster_labels + 1]
cluster_indices = np.where(km_clf != i)[0]
gene_expression_cluster = np.delete(
gene_expression_top_var_header_rows,
cluster_indices,
axis=0)
gene_headers_row_cluster = np.delete(
gene_expression_top_var_header_rows,
cluster_indices,
axis=0)
clfs_results[n_clusters].append(
(gene_headers_row_cluster, gene_headers_row_cluster))
desc = "k={} clustering cluster {} has {} genes".format(
n_clusters, i, len(gene_expression_cluster))
gene_list = ",".join(gene_headers_row_cluster)
url = check_enrichment(gene_list)
go_terms = []
uncorrectd_pvals = []
FDRs = []
go_names = []
go_ns = []
if calc_go:
g_res = g.run_study([
int(cur) for cur in ensembl2entrez_convertor(
gene_headers_row_cluster)
])
GO_results = [(cur.NS, cur.GO, cur.goterm.name,
cur.p_uncorrected, cur.p_fdr_bh)
for cur in g_res if cur.p_fdr_bh <= 0.05]
if len(GO_results) > 0:
go_ns, go_terms, go_names, uncorrectd_pvals, FDRs = zip(
*GO_results)
if len(enrichment_lists) != 0:
for j, cur in enumerate(enrichment_lists):
go_terms.append(
enrichment_list_file_names[j].split(".")[0])
uncorrectd_pvals.append(
calc_HG_test(
[x.split(".")[0] for x in tested_gene_list],
[x.split(".")[0] for x in cur], [
x.split(".")[0]
for x in gene_headers_row_cluster
]))
FDRs.append(".")
go_names.append(".")
go_ns.append(".")
output_rows.append((desc, "\r\n".join([
x.split(".")[0] for x in gene_headers_row_cluster
]), url, "\r\n".join(go_ns), "\r\n".join(go_terms),
"\r\n".join(go_names),
"\r\n".join(map(str, uncorrectd_pvals)),
"\r\n".join(map(str, FDRs))))
gene_sorted_heatmap = np.rot90(np.flip(
gene_expression_top_var[cluster_labels.argsort(), :], 1),
k=-1,
axes=(1, 0))
find_clusters(end_k,
gene_sorted_heatmap,
gene_expression_top_var_header_columns,
start_k,
e2g_convertor(gene_expression_top_var_header_rows),
tested_gene_list_file_name,
labels_assignment=labels_assignment_patients)
plot_heatmap(gene_expression_top_var,
gene_expression_top_var_header_columns,
labels_assignment,
gene_expression_top_var_header_rows,
tested_gene_list_file_name,
n_clusters=None,
label_index=None,
phenotype_heatmap=None)
gene_sorted_heatmap = np.rot90(np.flip(gene_expression_top_var, 1),
k=-1,
axes=(1, 0))
if cluster_algorithm == "hierarchical":
df = pd.DataFrame(data=gene_sorted_heatmap,
index=gene_expression_top_var_header_columns,
columns=gene_expression_top_var_header_rows)
# correlations = df.corr()
# correlations_array = np.asarray(df.corr())
#
# row_linkage = hierarchy.linkage(
# distance.pdist(correlations_array), method='average')
#
# col_linkage = hierarchy.linkage(
# distance.pdist(correlations_array.T), method='average')
# enrichment_gene_list = load_gene_list("uvm_mito_part.txt")
dct = dict(zip(np.unique(labels_assignment_patients[0]), "rbg"))
row_colors = map(dct.get, labels_assignment_patients[0])
dct = {1: 'b', 2: 'r'}
gene_expression_top_var_header_rows_trimmed = [
x.split(".")[0] for x in gene_expression_top_var_header_rows
]
# col_colors = map(dct.get, [2 if x in enrichment_gene_list else 1 for x in gene_expression_top_var_header_rows_trimmed])
g = sns.clustermap(df,
row_colors=row_colors,
metric="euclidean",
robust=True,
method="single")
# den_patients = scipy.cluster.hierarchy.dendrogram(g.dendrogram_row.linkage,
# labels=df.index,
# color_threshold=0.60)
den_genes = scipy.cluster.hierarchy.dendrogram(
g.dendrogram_col.linkage, labels=df.columns, color_threshold=0.7)
clusters = get_cluster_classes(den_genes)
g.savefig(
os.path.join(constants.BASE_PROFILE, "output",
"hierarchical_cluster_{}.png".format(time.time())))
for cur_labels_assignment_patient in labels_assignment_patients:
plot_heatmap(gene_sorted_heatmap,
gene_expression_top_var_header_rows,
[cur_labels_assignment_patient],
gene_expression_top_var_header_columns,
tested_gene_list_file_name,
n_clusters=None,
label_index=None,
phenotype_heatmap=None)
print_to_excel(
output_rows=output_rows,
gene_list_file_name=tested_gene_list_file_name.split(".")[0],
gene_expression_file_name=gene_expression_file_name.split(".")[0],
var_th_index=var_th_index)
