def findOverlapCoords(coordTree, coord_list, this_coord):
this_tree = coordTree
try:
this_key = this_tree.getKey()
except:
return
thisIsOverlap = False
if coordsOverlap(this_coord[0], this_coord[1], this_key[0], this_key[1]):
coord_list.append(this_tree.getKey())
thisIsOverlap = True
if not this_tree.getIsLeaf():
if this_coord < this_key or thisIsOverlap:
if this_tree.getLeft().isEmpty:
return
return findOverlapCoords(this_tree.getLeft(), coord_list,
this_coord)
if this_coord > this_key or thisIsOverlap:
if this_tree.getRight().isEmpty:
return
return findOverlapCoords(this_tree.getRight(), coord_list,
this_coord)
def hasOverlap(coordTree, this_coord):
this_tree = coordTree
try:
# Make sure that the tree is nonEmpty
tree_coord = this_tree.getKey()
except:
return False
if coordsOverlap(this_coord[0], this_coord[1], tree_coord[0],
tree_coord[1]):
return True
if not this_tree.getIsLeaf():
if this_coord < this_tree.getKey():
if this_tree.getLeft().isEmpty:
return False
return hasOverlap(this_tree.getLeft(), this_coord)
if this_coord > this_tree.getKey():
if this_tree.getRight().isEmpty:
return False
return hasOverlap(this_tree.getRight(), this_coord)
return False
def findOverlapCoords(coordTree, coord_list, this_coord):
this_tree = coordTree
try:
this_key = this_tree.getKey()
except:
return
thisIsOverlap = False
if coordsOverlap(this_coord[0],
this_coord[1],
this_key[0],
this_key[1]):
coord_list.append(this_tree.getKey())
thisIsOverlap = True
if not this_tree.getIsLeaf():
if this_coord < this_key or thisIsOverlap:
if this_tree.getLeft().isEmpty:
return
return findOverlapCoords(this_tree.getLeft(), coord_list, this_coord)
if this_coord > this_key or thisIsOverlap:
if this_tree.getRight().isEmpty:
return
return findOverlapCoords(this_tree.getRight(), coord_list, this_coord)
def hasOverlap(coordTree, this_coord):
this_tree = coordTree
try:
# Make sure that the tree is nonEmpty
tree_coord = this_tree.getKey()
except:
return False
if coordsOverlap(this_coord[0],
this_coord[1],
tree_coord[0],
tree_coord[1]):
return True
if not this_tree.getIsLeaf():
if this_coord < this_tree.getKey():
if this_tree.getLeft().isEmpty:
return False
return hasOverlap(this_tree.getLeft(), this_coord)
if this_coord > this_tree.getKey():
if this_tree.getRight().isEmpty:
return False
return hasOverlap(this_tree.getRight(), this_coord)
return False
def mergeRedundantEvents(as_type2chr2strand2redundantGroup2event):
"""
Because of the construction of the dictionary, some of the redundant
groups are overlapping.
"""
for as_type in as_type2chr2strand2redundantGroup2event:
for chr in as_type2chr2strand2redundantGroup2event[as_type]:
for strand in as_type2chr2strand2redundantGroup2event[as_type][
chr]:
for first_group in as_type2chr2strand2redundantGroup2event[
as_type][chr][strand]:
for second_group in as_type2chr2strand2redundantGroup2event[
as_type][chr][strand]:
if first_group == second_group:
continue
if coordsOverlap(first_group[2], first_group[3],
second_group[2], second_group[3]):
# Add sets together
new_set = as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][first_group].union(
as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][second_group])
new_start = min(first_group[2], second_group[2])
new_end = max(first_group[3], second_group[3])
new_region = (first_group[0], first_group[1],
new_start, new_end)
if new_region in as_type2chr2strand2redundantGroup2event[
as_type][chr][strand]:
as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][new_region].update(
new_set)
else:
as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][new_region] = new_set
# Remove old sets and return with flag
if new_region != first_group:
del as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][first_group]
if new_region != second_group:
del as_type2chr2strand2redundantGroup2event[
as_type][chr][strand][second_group]
return True
# If not returned from within the loop, then no change occurred
return False
def updateRedundantDictionary(as_type2chr2strand2redundantGroup2event, as_type, redundantRegion, event):
this_chr = redundantRegion[0]
this_strand = redundantRegion[1]
if as_type in as_type2chr2strand2redundantGroup2event:
if not this_chr in as_type2chr2strand2redundantGroup2event[as_type]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr] = {this_strand: {redundantRegion: set([event])}}
return
if not this_strand in as_type2chr2strand2redundantGroup2event[as_type][this_chr]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand] = {redundantRegion: set([event])}
return
foundOverlap = False
for redundantGroup in as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand]:
if coordsOverlap(redundantGroup[2], redundantGroup[3], redundantRegion[2], redundantRegion[3]):
foundOverlap = True
# Pick longest region as the key
region_start = min(redundantGroup[2], redundantRegion[2])
region_end = max(redundantGroup[3], redundantRegion[3])
# Copy the existing set of exon_names
cur_set = set(as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand][redundantGroup])
# Add the current event
cur_set.add(event)
del as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand][redundantGroup]
# Check for existing new group
new_group = (redundantGroup[0], redundantGroup[1], region_start, region_end)
if new_group in as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand][new_group].update(cur_set)
else:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand][new_group] = cur_set
break
if not foundOverlap:
# Add this non overlapping region on its own
as_type2chr2strand2redundantGroup2event[as_type][this_chr][this_strand][redundantRegion] = set([event])
else:
as_type2chr2strand2redundantGroup2event[as_type] = {this_chr: {this_strand: {redundantRegion: set([event])}}}
def mergeRedundantEvents(as_type2chr2strand2redundantGroup2event):
"""
Because of the construction of the dictionary, some of the redundant
groups are overlapping.
"""
for as_type in as_type2chr2strand2redundantGroup2event:
for chr in as_type2chr2strand2redundantGroup2event[as_type]:
for strand in as_type2chr2strand2redundantGroup2event[as_type][chr]:
for first_group in as_type2chr2strand2redundantGroup2event[as_type][chr][strand]:
for second_group in as_type2chr2strand2redundantGroup2event[as_type][chr][strand]:
if first_group == second_group:
continue
if coordsOverlap(first_group[2], first_group[3], second_group[2], second_group[3]):
# Add sets together
new_set = as_type2chr2strand2redundantGroup2event[as_type][chr][strand][first_group].union(
as_type2chr2strand2redundantGroup2event[as_type][chr][strand][second_group]
)
new_start = min(first_group[2], second_group[2])
new_end = max(first_group[3], second_group[3])
new_region = (first_group[0], first_group[1], new_start, new_end)
if new_region in as_type2chr2strand2redundantGroup2event[as_type][chr][strand]:
as_type2chr2strand2redundantGroup2event[as_type][chr][strand][new_region].update(
new_set
)
else:
as_type2chr2strand2redundantGroup2event[as_type][chr][strand][new_region] = new_set
# Remove old sets and return with flag
if new_region != first_group:
del as_type2chr2strand2redundantGroup2event[as_type][chr][strand][first_group]
if new_region != second_group:
del as_type2chr2strand2redundantGroup2event[as_type][chr][strand][second_group]
return True
# If not returned from within the loop, then no change occurred
return False
def updateRedundantDictionary(as_type2chr2strand2redundantGroup2event, as_type,
redundantRegion, event):
this_chr = redundantRegion[0]
this_strand = redundantRegion[1]
if as_type in as_type2chr2strand2redundantGroup2event:
if not this_chr in as_type2chr2strand2redundantGroup2event[as_type]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr] = {
this_strand: {
redundantRegion: set([event])
}
}
return
if not this_strand in as_type2chr2strand2redundantGroup2event[as_type][
this_chr]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][
this_strand] = {
redundantRegion: set([event])
}
return
foundOverlap = False
for redundantGroup in as_type2chr2strand2redundantGroup2event[as_type][
this_chr][this_strand]:
if coordsOverlap(redundantGroup[2], redundantGroup[3],
redundantRegion[2], redundantRegion[3]):
foundOverlap = True
# Pick longest region as the key
region_start = min(redundantGroup[2], redundantRegion[2])
region_end = max(redundantGroup[3], redundantRegion[3])
# Copy the existing set of exon_names
cur_set = set(as_type2chr2strand2redundantGroup2event[as_type]
[this_chr][this_strand][redundantGroup])
# Add the current event
cur_set.add(event)
del as_type2chr2strand2redundantGroup2event[as_type][this_chr][
this_strand][redundantGroup]
# Check for existing new group
new_group = (redundantGroup[0], redundantGroup[1],
region_start, region_end)
if new_group in as_type2chr2strand2redundantGroup2event[
as_type][this_chr][this_strand]:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][
this_strand][new_group].update(cur_set)
else:
as_type2chr2strand2redundantGroup2event[as_type][this_chr][
this_strand][new_group] = cur_set
break
if not foundOverlap:
# Add this non overlapping region on its own
as_type2chr2strand2redundantGroup2event[as_type][this_chr][
this_strand][redundantRegion] = set([event])
else:
as_type2chr2strand2redundantGroup2event[as_type] = {
this_chr: {
this_strand: {
redundantRegion: set([event])
}
}
}