def create_file(self):
# Expanding summits
tfbs_summit_regions = GenomicRegionSet("TFBS Summit Regions")
tfbs_summit_regions.read_bed(self.tfbs_summit_fname)
for region in iter(tfbs_summit_regions):
summit = int(region.data.split()[-1]) + region.initial
region.initial = max(summit - (self.peak_ext / 2), 0)
region.final = summit + (self.peak_ext / 2)
# Calculating intersections
mpbs_regions = GenomicRegionSet("MPBS Regions")
mpbs_regions.read_bed(self.mpbs_fname)
tfbs_summit_regions.sort()
mpbs_regions.sort()
with_overlap_regions = mpbs_regions.intersect(tfbs_summit_regions, mode=OverlapType.ORIGINAL)
without_overlap_regions = mpbs_regions.subtract(tfbs_summit_regions, whole_region=True)
tfbs_regions = GenomicRegionSet("TFBS Regions")
for region in iter(with_overlap_regions):
region.name = region.name.split(":")[0] + ":Y"
tfbs_regions.add(region)
for region in iter(without_overlap_regions):
region.name = region.name.split(":")[0] + ":N"
tfbs_regions.add(region)
tfbs_regions.sort()
tfbs_fname = os.path.join(self.output_location, "{}.bed".format(self.mpbs_name))
tfbs_regions.write_bed(tfbs_fname)
def subtract(self, x):
"""
Subtract GenomicVariantSet.
*Keyword arguments:*
- x -- instance of GenomicVariantSet which is subtracted
"""
tmp = GenomicRegionSet.subtract(self, x, whole_region=False)
self.sequences = self._reconstruct_info(tmp)
def test_subtract_exact(self):
reference = GenomicRegionSet("reference")
reference.read(os.path.join(os.path.dirname(__file__), "test_result.bed"))
background = GenomicRegionSet("background")
background.read(os.path.join(os.path.dirname(__file__), "test_background.bed"))
target = GenomicRegionSet("target")
target.read(os.path.join(os.path.dirname(__file__), "test_target.bed"))
background_tmp = background.subtract(target, exact=True)
reference.sort()
self.assertEqual(len(background_tmp.sequences), len(reference.sequences))
for region, region_ref in zip(background_tmp.sequences, reference.sequences):
self.assertEqual(region.__cmp__(region_ref), 0)
def test_subtract_exact(self):
reference = GenomicRegionSet("reference")
reference.read(
os.path.join(os.path.dirname(__file__), "test_result.bed"))
background = GenomicRegionSet("background")
background.read(
os.path.join(os.path.dirname(__file__), "test_background.bed"))
target = GenomicRegionSet("target")
target.read(os.path.join(os.path.dirname(__file__), "test_target.bed"))
background_tmp = background.subtract(target, exact=True)
reference.sort()
self.assertEqual(len(background_tmp.sequences),
len(reference.sequences))
for region, region_ref in zip(background_tmp.sequences,
reference.sequences):
self.assertEqual(region, region_ref)
def create_file(self):
# Expanding summits
tfbs_summit_regions = GenomicRegionSet("TFBS Summit Regions")
tfbs_summit_regions.read_bed(self.tfbs_summit_fname)
for region in iter(tfbs_summit_regions):
summit = int(region.data.split()[-1]) + region.initial
region.initial = max(summit - (self.peak_ext / 2), 0)
region.final = summit + (self.peak_ext / 2)
# Calculating intersections
mpbs_regions = GenomicRegionSet("MPBS Regions")
mpbs_regions.read_bed(self.mpbs_fname)
tfbs_summit_regions.sort()
mpbs_regions.sort()
with_overlap_regions = mpbs_regions.intersect(
tfbs_summit_regions, mode=OverlapType.ORIGINAL)
without_overlap_regions = mpbs_regions.subtract(tfbs_summit_regions,
whole_region=True)
tfbs_regions = GenomicRegionSet("TFBS Regions")
for region in iter(with_overlap_regions):
region.name = region.name.split(":")[0] + ":Y"
tfbs_regions.add(region)
for region in iter(without_overlap_regions):
region.name = region.name.split(":")[0] + ":N"
tfbs_regions.add(region)
tfbs_regions.sort()
tfbs_fname = os.path.join(self.output_location,
"{}.bed".format(self.mpbs_name))
tfbs_regions.write_bed(tfbs_fname)
def chip_evaluate(args):
# Evaluate Statistics
fpr = dict()
tpr = dict()
roc_auc_1 = dict()
roc_auc_10 = dict()
roc_auc_50 = dict()
roc_auc_100 = dict()
recall = dict()
precision = dict()
prc_auc_1 = dict()
prc_auc_10 = dict()
prc_auc_50 = dict()
prc_auc_100 = dict()
footprint_file = args.footprint_file.split(",")
footprint_name = args.footprint_name.split(",")
footprint_type = args.footprint_type.split(",")
max_score = 0
if "SEG" in footprint_type:
mpbs_regions = GenomicRegionSet("TFBS")
mpbs_regions.read(args.tfbs_file)
# Verifying the maximum score of the MPBS file
for region in iter(mpbs_regions):
score = int(region.data.split("\t")[0])
if score > max_score:
max_score = score
max_score += 1
for i in range(len(footprint_file)):
footprints_regions = GenomicRegionSet("Footprints Prediction")
footprints_regions.read(footprint_file[i])
footprints_regions.sort()
if footprint_type[i] == "SEG":
# Increasing the score of MPBS entry once if any overlaps found in the predicted footprints.
increased_score_mpbs_regions = GenomicRegionSet("Increased Regions")
intersect_regions = mpbs_regions.intersect(footprints_regions, mode=OverlapType.ORIGINAL)
for region in iter(intersect_regions):
region.data = str(int(region.data.split("\t")[0]) + max_score)
increased_score_mpbs_regions.add(region)
# Keep the score of remained MPBS entry unchanged
without_intersect_regions = mpbs_regions.subtract(footprints_regions, whole_region=True)
for region in iter(without_intersect_regions):
increased_score_mpbs_regions.add(region)
increased_score_mpbs_regions.sort_score()
fpr[i], tpr[i], roc_auc_1[i], roc_auc_10[i], roc_auc_50[i], roc_auc_100[i] = \
roc_curve(increased_score_mpbs_regions)
recall[i], precision[i], prc_auc_1[i], prc_auc_10[i], prc_auc_50[i], prc_auc_100[i] = \
precision_recall_curve(increased_score_mpbs_regions)
elif footprint_type[i] == "SC":
footprints_regions.sort_score()
fpr[i], tpr[i], roc_auc_1[i], roc_auc_10[i], roc_auc_50[i], roc_auc_100[i] = \
roc_curve(footprints_regions)
recall[i], precision[i], prc_auc_1[i], prc_auc_10[i], prc_auc_50[i], prc_auc_100[i] = \
precision_recall_curve(footprints_regions)
# Output the statistics results into text
stats_fname = os.path.join(args.output_location, "{}_stats.txt".format(args.output_prefix))
stats_header = ["METHOD", "AUC_100", "AUC_50", "AUC_10", "AUC_1", "AUPR_100", "AUPR_50", "AUPR_10", "AUPR_1"]
with open(stats_fname, "w") as stats_file:
stats_file.write("\t".join(stats_header) + "\n")
for i in range(len(footprint_name)):
stats_file.write(footprint_name[i] + "\t" +
str(roc_auc_100[i]) + "\t" + str(roc_auc_50[i]) + "\t" + str(roc_auc_10[i]) + "\t" +
str(roc_auc_1[i]) + "\t" + str(prc_auc_100[i]) + "\t" + str(prc_auc_50[i]) + "\t" +
str(prc_auc_10[i]) + "\t" + str(prc_auc_1[i]) + "\n")
# Output the curves
if args.print_roc_curve:
label_x = "False Positive Rate"
label_y = "True Positive Rate"
curve_name = "ROC"
plot_curve(footprint_name, args.output_location, fpr, tpr, roc_auc_100, label_x, label_y, args.output_prefix,
curve_name)
if args.print_pr_curve:
label_x = "Recall"
label_y = "Precision"
curve_name = "PRC"
plot_curve(footprint_name, args.output_location, recall, precision, prc_auc_100, label_x, label_y,
args.output_prefix, curve_name)
output_points(footprint_name, args.output_location, args.output_prefix, fpr, tpr, recall, precision)
def chip_evaluate(self):
"""
This evaluation methodology uses motif-predicted binding sites (MPBSs) together with TF ChIP-seq data
to evaluate the footprint predictions.
return:
"""
# Evaluate Statistics
fpr = dict()
tpr = dict()
roc_auc = dict()
roc_auc_1 = dict()
roc_auc_2 = dict()
recall = dict()
precision = dict()
prc_auc = dict()
if "SEG" in self.footprint_type:
mpbs_regions = GenomicRegionSet("TFBS")
mpbs_regions.read_bed(self.tfbs_file)
mpbs_regions.sort()
# Verifying the maximum score of the MPBS file
max_score = -99999999
for region in iter(mpbs_regions):
score = int(region.data)
if score > max_score:
max_score = score
max_score += 1
for i in range(len(self.footprint_file)):
footprints_regions = GenomicRegionSet("Footprints Prediction")
footprints_regions.read_bed(self.footprint_file[i])
# Sort footprint prediction bed files
footprints_regions.sort()
if self.footprint_type[i] == "SEG":
# Increasing the score of MPBS entry once if any overlaps found in the predicted footprints.
increased_score_mpbs_regions = GenomicRegionSet("Increased Regions")
intersect_regions = mpbs_regions.intersect(footprints_regions, mode=OverlapType.ORIGINAL)
for region in iter(intersect_regions):
region.data = str(int(region.data) + max_score)
increased_score_mpbs_regions.add(region)
# Keep the score of remained MPBS entry unchanged
without_intersect_regions = mpbs_regions.subtract(footprints_regions, whole_region=True)
for region in iter(without_intersect_regions):
increased_score_mpbs_regions.add(region)
increased_score_mpbs_regions.sort_score()
fpr[i], tpr[i], roc_auc[i], roc_auc_1[i], roc_auc_2[i] = self.roc_curve(increased_score_mpbs_regions)
recall[i], precision[i], prc_auc[i] = self.precision_recall_curve(increased_score_mpbs_regions)
elif self.footprint_type[i] == "SC":
footprints_regions.sort_score()
fpr[i], tpr[i], roc_auc[i], roc_auc_1[i], roc_auc_2[i] = self.roc_curve(footprints_regions)
recall[i], precision[i], prc_auc[i] = self.precision_recall_curve(footprints_regions)
# Output the statistics results into text
stats_fname = self.output_location + self.tf_name + "_stats.txt"
stats_header = ["METHOD", "AUC_100", "AUC_10", "AUC_1", "AUPR"]
with open(stats_fname, "w") as stats_file:
stats_file.write("\t".join(stats_header) + "\n")
for i in range(len(self.footprint_name)):
stats_file.write(self.footprint_name[i] + "\t" + str(roc_auc[i]) + "\t" + str(roc_auc_1[i]) + "\t"
+ str(roc_auc_2[i]) + "\t" + str(prc_auc[i]) + "\n")
# Output the curves
if self.print_roc_curve:
label_x = "False Positive Rate"
label_y = "True Positive Rate"
curve_name = "ROC"
self.plot_curve(fpr, tpr, roc_auc, label_x, label_y, self.tf_name, curve_name)
if self.print_pr_curve:
label_x = "Recall"
label_y = "Precision"
curve_name = "PRC"
self.plot_curve(recall, precision, prc_auc, label_x, label_y, self.tf_name, curve_name)
self.output_points(self.tf_name, fpr, tpr, recall, precision)
def chip_evaluate(args):
# Evaluate Statistics
fpr = dict()
tpr = dict()
roc_auc_1 = dict()
roc_auc_10 = dict()
roc_auc_50 = dict()
roc_auc_100 = dict()
recall = dict()
precision = dict()
prc_auc_1 = dict()
prc_auc_10 = dict()
prc_auc_50 = dict()
prc_auc_100 = dict()
footprint_file = args.footprint_file.split(",")
footprint_name = args.footprint_name.split(",")
footprint_type = args.footprint_type.split(",")
max_score = 0
if "SEG" in footprint_type:
mpbs_regions = GenomicRegionSet("TFBS")
mpbs_regions.read(args.tfbs_file)
# Verifying the maximum score of the MPBS file
for region in iter(mpbs_regions):
score = int(region.data.split("\t")[0])
if score > max_score:
max_score = score
max_score += 1
max_points = []
for i in range(len(footprint_file)):
footprints_regions = GenomicRegionSet("Footprints Prediction")
footprints_regions.read(footprint_file[i])
footprints_regions.sort()
if footprint_type[i] == "SEG":
# Increasing the score of MPBS entry once if any overlaps found in the predicted footprints.
increased_score_mpbs_regions = GenomicRegionSet("Increased Regions")
intersect_regions = mpbs_regions.intersect(footprints_regions, mode=OverlapType.ORIGINAL)
for region in iter(intersect_regions):
region.data = str(int(region.data.split("\t")[0]) + max_score)
increased_score_mpbs_regions.add(region)
# Keep the score of remained MPBS entry unchanged
without_intersect_regions = mpbs_regions.subtract(footprints_regions, whole_region=True)
for region in iter(without_intersect_regions):
increased_score_mpbs_regions.add(region)
increased_score_mpbs_regions.sort_score()
fpr[i], tpr[i], roc_auc_1[i], roc_auc_10[i], roc_auc_50[i], roc_auc_100[i] = \
roc_curve(increased_score_mpbs_regions)
recall[i], precision[i], prc_auc_1[i], prc_auc_10[i], prc_auc_50[i], prc_auc_100[i] = \
precision_recall_curve(increased_score_mpbs_regions)
max_points.append(len(intersect_regions))
elif footprint_type[i] == "SC":
footprints_regions.sort_score()
fpr[i], tpr[i], roc_auc_1[i], roc_auc_10[i], roc_auc_50[i], roc_auc_100[i] = \
roc_curve(footprints_regions)
recall[i], precision[i], prc_auc_1[i], prc_auc_10[i], prc_auc_50[i], prc_auc_100[i] = \
precision_recall_curve(footprints_regions)
max_points.append(len(footprints_regions))
# Output the statistics results into text
stats_fname = os.path.join(args.output_location, "{}_stats.txt".format(args.output_prefix))
stats_header = ["METHOD", "AUC_100", "AUC_50", "AUC_10", "AUC_1", "AUPR_100", "AUPR_50", "AUPR_10", "AUPR_1"]
with open(stats_fname, "w") as stats_file:
stats_file.write("\t".join(stats_header) + "\n")
for i in range(len(footprint_name)):
stats_file.write(footprint_name[i] + "\t" +
str(roc_auc_100[i]) + "\t" + str(roc_auc_50[i]) + "\t" + str(roc_auc_10[i]) + "\t" +
str(roc_auc_1[i]) + "\t" + str(prc_auc_100[i]) + "\t" + str(prc_auc_50[i]) + "\t" +
str(prc_auc_10[i]) + "\t" + str(prc_auc_1[i]) + "\n")
# Output the curves
if args.print_roc_curve:
label_x = "False Positive Rate"
label_y = "True Positive Rate"
curve_name = "ROC"
plot_curve(footprint_name, args.output_location, fpr, tpr, roc_auc_100, label_x, label_y, args.output_prefix,
curve_name, max_points=max_points)
if args.print_pr_curve:
label_x = "Recall"
label_y = "Precision"
curve_name = "PRC"
plot_curve(footprint_name, args.output_location, recall, precision, prc_auc_100, label_x, label_y,
args.output_prefix, curve_name, max_points=max_points)
output_points(footprint_name, args.output_location, args.output_prefix, fpr, tpr, recall, precision)
def chip_evaluate(self):
"""
This evaluation methodology uses motif-predicted binding sites (MPBSs) together with TF ChIP-seq data
to evaluate the footprint predictions.
return:
"""
# Evaluate Statistics
fpr = dict()
tpr = dict()
roc_auc = dict()
roc_auc_1 = dict()
roc_auc_2 = dict()
recall = dict()
precision = dict()
prc_auc = dict()
if "SEG" in self.footprint_type:
mpbs_regions = GenomicRegionSet("TFBS")
mpbs_regions.read_bed(self.tfbs_file)
mpbs_regions.sort()
# Verifying the maximum score of the MPBS file
max_score = -99999999
for region in iter(mpbs_regions):
score = int(region.data)
if score > max_score:
max_score = score
max_score += 1
for i in range(len(self.footprint_file)):
footprints_regions = GenomicRegionSet("Footprints Prediction")
footprints_regions.read_bed(self.footprint_file[i])
# Sort footprint prediction bed files
footprints_regions.sort()
if self.footprint_type[i] == "SEG":
# Increasing the score of MPBS entry once if any overlaps found in the predicted footprints.
increased_score_mpbs_regions = GenomicRegionSet(
"Increased Regions")
intersect_regions = mpbs_regions.intersect(
footprints_regions, mode=OverlapType.ORIGINAL)
for region in iter(intersect_regions):
region.data = str(int(region.data) + max_score)
increased_score_mpbs_regions.add(region)
# Keep the score of remained MPBS entry unchanged
without_intersect_regions = mpbs_regions.subtract(
footprints_regions, whole_region=True)
for region in iter(without_intersect_regions):
increased_score_mpbs_regions.add(region)
increased_score_mpbs_regions.sort_score()
fpr[i], tpr[i], roc_auc[i], roc_auc_1[i], roc_auc_2[
i] = self.roc_curve(increased_score_mpbs_regions)
recall[i], precision[i], prc_auc[
i] = self.precision_recall_curve(
increased_score_mpbs_regions)
elif self.footprint_type[i] == "SC":
footprints_regions.sort_score()
fpr[i], tpr[i], roc_auc[i], roc_auc_1[i], roc_auc_2[
i] = self.roc_curve(footprints_regions)
recall[i], precision[i], prc_auc[
i] = self.precision_recall_curve(footprints_regions)
# Output the statistics results into text
stats_fname = self.output_location + self.tf_name + "_stats.txt"
stats_header = ["METHOD", "AUC_100", "AUC_10", "AUC_1", "AUPR"]
with open(stats_fname, "w") as stats_file:
stats_file.write("\t".join(stats_header) + "\n")
for i in range(len(self.footprint_name)):
stats_file.write(self.footprint_name[i] + "\t" +
str(roc_auc[i]) + "\t" + str(roc_auc_1[i]) +
"\t" + str(roc_auc_2[i]) + "\t" +
str(prc_auc[i]) + "\n")
# Output the curves
if self.print_roc_curve:
label_x = "False Positive Rate"
label_y = "True Positive Rate"
curve_name = "ROC"
self.plot_curve(fpr, tpr, roc_auc, label_x, label_y, self.tf_name,
curve_name)
if self.print_pr_curve:
label_x = "Recall"
label_y = "Precision"
curve_name = "PRC"
self.plot_curve(recall, precision, prc_auc, label_x, label_y,
self.tf_name, curve_name)
self.output_points(self.tf_name, fpr, tpr, recall, precision)
print("output:\t" + args.o)
if not args.t:
print("Please define the file for target regions.")
sys.exit(1)
else:
print("target:\t" + args.t)
# with open(args.target) as f:
t = GenomicRegionSet("targets")
t.read_bed(args.t)
# with open(args.i) as fi, open(args.o, "w") as fo:
input_regions = GenomicRegionSet("input")
input_regions.read_bed(args.i)
output_regions = input_regions.subtract(t,whole_region=True)
output_regions.write_bed(args.o)
print("complete.")
############### BED add columns #############################################
elif args.mode == "bed_add_columns":
print("input:\t" + args.i)
print("reference:\t" + args.ref)
print("output:\t" + args.o)
if not args.ref:
print("Please define the file for reference.")
sys.exit(1)
with open(args.ref) as f:
genes = {}
def subtract(self, x):
tmp = GenomicRegionSet.subtract(self, x, whole_region=False)
self.sequences = self._reconstruct_info(tmp)
def subtract(self, x):
tmp = GenomicRegionSet.subtract(self, x, whole_region=False)
self.sequences = self._reconstruct_info(tmp)
