def annotate_junctions(junctions, reference_genepred_filename, splicebases):
annot_struct = genepred_basics.get_gene_annotation_data_structure(reference_genepred_filename)
genepred = genepred_basics.get_per_chromosome_array(reference_genepred_filename)
result = {}
for junc_name in junctions:
splice = splicebases[junc_name]
junc = junctions[junc_name]
genepred_entries = []
if junc["chr1"] in genepred:
genepred_entries = genepred[junc["chr1"]]
leftend = "N" # are we on the end of an exon on the left side
if junc["dir1"] == "+" and genepred_basics.is_exon_end(junc["chr1"], int(junc["coo1"]), genepred_entries):
leftend = "Y"
elif junc["dir1"] == "-" and genepred_basics.is_exon_start(junc["chr1"], int(junc["coo1"]), genepred_entries):
leftend = "Y"
genepred_entries = []
if junc["chr2"] in genepred:
genepred_entries = genepred[junc["chr2"]]
rightend = "N"
if junc["dir2"] == "+" and genepred_basics.is_exon_start(junc["chr2"], int(junc["coo2"]), genepred_entries):
rightend = "Y"
elif junc["dir2"] == "-" and genepred_basics.is_exon_end(junc["chr2"], int(junc["coo2"]), genepred_entries):
rightend = "Y"
annot1 = genepred_basics.gene_annotate_by_coordinates(
junc["chr1"], int(junc["coo1"]) - 1, int(junc["coo1"]), annot_struct
)
annot2 = genepred_basics.gene_annotate_by_coordinates(
junc["chr2"], int(junc["coo2"]) - 1, int(junc["coo2"]), annot_struct
)
best1 = junc["chr1"] + ":" + str(junc["coo1"])
best1dir = ""
best2 = junc["chr2"] + ":" + str(junc["coo2"])
best2dir = ""
name1 = ""
name2 = ""
if annot1:
best1 = annot1[0][0]
name1 = annot1[0][0]
best1dir = annot1[0][1]
if annot2:
best2 = annot2[0][0]
name2 = annot2[0][0]
best2dir = annot2[0][1]
evidence_transcript = "N"
evidence_splice = "N"
action = "unknown"
if best1dir == "+" and best2dir == "+":
if junc["dir1"] == "-" and junc["dir2"] == "-":
action = "reverse"
elif junc["dir1"] == "+" and junc["dir2"] == "+":
action = "no reverse"
else:
action = "unknown"
elif best1dir == "-" and best2dir == "-":
if junc["dir1"] == "+" and junc["dir2"] == "+":
action = "reverse"
elif junc["dir1"] == "-" and junc["dir2"] == "-":
action = "no reverse"
else:
action = "unknown"
if best1dir == "+" and best2dir == "-":
if junc["dir1"] == "-" and junc["dir2"] == "+":
action = "reverse"
elif junc["dir1"] == "+" and junc["dir2"] == "-":
action = "no reverse"
else:
action = "unknown"
if best1dir == "-" and best2dir == "+":
if junc["dir1"] == "+" and junc["dir2"] == "-":
action = "reverse"
elif junc["dir1"] == "-" and junc["dir2"] == "+":
action = "no reverse"
else:
action = "unknown"
if best1dir == "-" and best2dir == "":
if junc["dir1"] == "+":
action = "reverse"
elif junc["dir1"] == "-":
action = "no reverse"
else:
action = "unknown"
if best1dir == "+" and best2dir == "":
if junc["dir1"] == "-":
action = "reverse"
elif junc["dir1"] == "+":
action = "no reverse"
else:
action = "unknown"
if best1dir == "" and best2dir == "+":
if junc["dir2"] == "-":
action = "reverse"
elif junc["dir2"] == "+":
action = "no reverse"
else:
action = "unknown"
if best1dir == "" and best2dir == "-":
if junc["dir2"] == "+":
action = "reverse"
elif junc["dir2"] == "-":
action = "no reverse"
else:
action = "unknown"
# set if we have transcript evidence
if action != "unknown":
evidence_transcript = "Y"
# set if we have splice evidence
if action == "no reverse" and splice == "GT AG":
evidence_splice = "Y"
if action == "reverse" and splice == "CT AC":
evidence_splice = "Y"
if action == "unknown" and (splice == "GT AG" or splice == "CT AC"):
evidence_splice = "Y"
# if action is still unknown, use splice site to make choice to reverse
if action == "unknown":
if splice == "GT AG":
action = "no reverse"
elif splice == "CT AC":
action = "reverse"
# switch our left and right end assignments if reverse
if action == "reverse":
temp1 = leftend
leftend = rightend
rightend = temp1
names = ""
if action == "no reverse":
names = (
junc["chr1"]
+ "\t"
+ junc["coo1"]
+ "\t"
+ junc["dir1"]
+ "\t"
+ leftend
+ "\t"
+ junc["chr2"]
+ "\t"
+ junc["coo2"]
+ "\t"
+ junc["dir2"]
+ "\t"
+ rightend
+ "\t"
+ best1
+ "-"
+ best2
+ "\t"
+ best1dir
+ "/"
+ best2dir
+ "\t"
+ evidence_transcript
+ "\t"
+ evidence_splice
+ "\t"
+ name1
+ "\t"
+ name2
)
elif action == "reverse":
names = (
junc["chr2"]
+ "\t"
+ junc["coo2"]
+ "\t"
+ opposite(junc["dir2"])
+ "\t"
+ leftend
+ "\t"
+ junc["chr1"]
+ "\t"
+ junc["coo1"]
+ "\t"
+ opposite(junc["dir1"])
+ "\t"
+ rightend
+ "\t"
+ best2
+ "-"
+ best1
+ "\t"
+ best2dir
+ "/"
+ best1dir
+ "\t"
+ evidence_transcript
+ "\t"
+ evidence_splice
+ "\t"
+ name2
+ "\t"
+ name1
)
else:
names = (
junc["chr1"]
+ "\t"
+ junc["coo1"]
+ "\t"
+ junc["dir1"]
+ "\t"
+ leftend
+ "\t"
+ junc["chr2"]
+ "\t"
+ junc["coo2"]
+ "\t"
+ junc["dir2"]
+ "\t"
+ rightend
+ "\t"
+ best1
+ "-"
+ best2
+ "\t"
+ best1dir
+ "/"
+ best2dir
+ "\t"
+ evidence_transcript
+ "\t"
+ evidence_splice
+ "\t"
+ name1
+ "\t"
+ name2
)
result[junc_name] = (
junc_name
+ "\t"
+ names
+ "\t"
+ str(junc["lr_count"])
+ "\t"
+ junc["type1nr"]
+ "\t"
+ junc["type1"]
+ "\t"
+ junc["type2"]
+ "\t"
+ junc["type3"]
+ "\t"
+ junc["type4"]
+ "\t"
+ str(junc["max_min_side_lengths"])
+ "\t"
+ str(junc["max_min_unique_side_counts"])
+ "\t"
+ junc["max_min_flanking_distance"]
)
return result
def annotate_junctions(junctions, reference_genepred_filename):
annot_struct = genepred_basics.get_gene_annotation_data_structure(
reference_genepred_filename)
genepred = genepred_basics.get_per_chromosome_array(
reference_genepred_filename)
for entry_set in junctions:
f = entry_set['entries']
d = entry_set['dictionary']
if d['Gene_1'] != '' and d['Gene_2'] != '':
print "\t".join(f) #already annotated. move on
continue
[best1dir, best2dir] = d['Transcript_strands'].split("/")
leftend = d[
'End_of_reference_exon_1'] # are we on the end of an exon on the left side
name1 = d['Gene_1']
best1 = name1
name1change = 0
if d['Gene_1'] == '':
best1 = d['Chromosome_1'] + ":" + str(d['Coordinate_1'])
genepred_entries = []
if d['Chromosome_1'] in genepred:
genepred_entries = genepred[d['Chromosome_1']]
if d['Direction_1'] == '+' and genepred_basics.is_exon_end(
d['Chromosome_1'], int(d['Coordinate_1']),
genepred_entries):
leftend = 'Y'
elif d['Direction_1'] == '-' and genepred_basics.is_exon_start(
d['Chromosome_1'], int(d['Coordinate_1']),
genepred_entries):
leftend = 'Y'
annot1 = genepred_basics.gene_annotate_by_coordinates(
d['Chromosome_1'],
int(d['Coordinate_1']) - 1, int(d['Coordinate_1']),
annot_struct)
if annot1:
name1change = 1
best1 = annot1[0][0]
name1 = annot1[0][0]
best1dir = annot1[0][1]
rightend = d['Start_of_reference_exon_2']
name2 = d['Gene_2']
best2 = name2
name2change = 0
if d['Gene_2'] == '':
best2 = d['Chromosome_2'] + ":" + str(d['Coordinate_2'])
genepred_entries = []
if d['Chromosome_2'] in genepred:
genepred_entries = genepred[d['Chromosome_2']]
if d['Direction_2'] == '+' and genepred_basics.is_exon_start(
d['Chromosome_2'], int(d['Coordinate_2']),
genepred_entries):
rightend = 'Y'
elif d['Direction_2'] == '-' and genepred_basics.is_exon_end(
d['Chromosome_2'], int(d['Coordinate_2']),
genepred_entries):
rightend = 'Y'
annot2 = genepred_basics.gene_annotate_by_coordinates(
d['Chromosome_2'],
int(d['Coordinate_2']) - 1, int(d['Coordinate_2']),
annot_struct)
if annot2:
name2change = 1
best2 = annot2[0][0]
name2 = annot2[0][0]
best2dir = annot2[0][1]
if name1change == 0 and name2change == 0:
print "\t".join(f)
continue # didn't find any additional name information
evidence_transcript = 'N'
evidence_splice = 'N'
action = 'unknown'
junc = {}
junc['dir1'] = d['Direction_1']
junc['dir2'] = d['Direction_2']
junc['coo1'] = d['Coordinate_1']
junc['coo2'] = d['Coordinate_2']
junc['chr1'] = d['Chromosome_1']
junc['chr2'] = d['Chromosome_2']
if best1dir == '+' and best2dir == '+':
if junc['dir1'] == '-' and junc['dir2'] == '-':
action = "reverse"
elif junc['dir1'] == '+' and junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
elif best1dir == '-' and best2dir == '-':
if junc['dir1'] == '+' and junc['dir2'] == '+':
action = "reverse"
elif junc['dir1'] == '-' and junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '+' and best2dir == '-':
if junc['dir1'] == '-' and junc['dir2'] == '+':
action = "reverse"
elif junc['dir1'] == '+' and junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '-' and best2dir == '+':
if junc['dir1'] == '+' and junc['dir2'] == '-':
action = "reverse"
elif junc['dir1'] == '-' and junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '-' and best2dir == '':
if junc['dir1'] == '+':
action = "reverse"
elif junc['dir1'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '+' and best2dir == '':
if junc['dir1'] == '-':
action = "reverse"
elif junc['dir1'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '' and best2dir == '+':
if junc['dir2'] == '-':
action = "reverse"
elif junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '' and best2dir == '-':
if junc['dir2'] == '+':
action = "reverse"
elif junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
#set if we have transcript evidence
if action != "unknown":
evidence_transcript = 'Y'
# switch our left and right end assignments if reverse
if action == "reverse":
temp1 = leftend
leftend = rightend
rightend = temp1
names = ''
if action == "no reverse":
names = junc['chr1'] + "\t" + str(
junc['coo1']
) + "\t" + junc['dir1'] + "\t" + leftend + "\t" + junc['chr2'] + "\t" + str(
junc['coo2']
) + "\t" + junc[
'dir2'] + "\t" + rightend + "\t" + best1 + "-" + best2 + "\t" + best1dir + "/" + best2dir + "\t" + evidence_transcript + "\t" + evidence_splice + "\t" + name1 + "\t" + name2
elif action == "reverse":
names = junc['chr2'] + "\t" + str(junc['coo2']) + "\t" + opposite(
junc['dir2']
) + "\t" + leftend + "\t" + junc['chr1'] + "\t" + str(
junc['coo1']
) + "\t" + opposite(
junc['dir1']
) + "\t" + rightend + "\t" + best2 + "-" + best1 + "\t" + best2dir + "/" + best1dir + "\t" + evidence_transcript + "\t" + evidence_splice + "\t" + name2 + "\t" + name1
else:
names = junc['chr1'] + "\t" + str(
junc['coo1']
) + "\t" + junc['dir1'] + "\t" + leftend + "\t" + junc['chr2'] + "\t" + str(
junc['coo2']
) + "\t" + junc[
'dir2'] + "\t" + rightend + "\t" + best1 + "-" + best2 + "\t" + best1dir + "/" + best2dir + "\t" + evidence_transcript + "\t" + evidence_splice + "\t" + name1 + "\t" + name2
print d['Alphabetized_junction_name'] + "\t" + names + "\t" + str(
d['LR_count']) + "\t" + str(d['nr_Type1_count']) + "\t" + str(
d['Type1_count']) + "\t" + str(d['Type2_count']) + "\t" + str(
d['Type3_count']) + "\t" + str(
d['Type4_count']) + "\t" + str(
d['Max_min_side_length']) + "\t" + str(
d['Max_min_side_unique']
) + "\t" + d['Max_min_side_flank']
def annotate_junctions(junctions,reference_genepred_filename, splicebases):
annot_struct = genepred_basics.get_gene_annotation_data_structure(reference_genepred_filename)
genepred = genepred_basics.get_per_chromosome_array(reference_genepred_filename)
result = {}
for junc_name in junctions:
splice = splicebases[junc_name]
junc = junctions[junc_name]
genepred_entries = []
if junc['chr1'] in genepred: genepred_entries = genepred[junc['chr1']]
leftend = 'N' # are we on the end of an exon on the left side
if junc['dir1'] == '+' and genepred_basics.is_exon_end(junc['chr1'],int(junc['coo1']),genepred_entries):
leftend = 'Y'
elif junc['dir1'] == '-' and genepred_basics.is_exon_start(junc['chr1'],int(junc['coo1']),genepred_entries):
leftend = 'Y'
genepred_entries = []
if junc['chr2'] in genepred: genepred_entries = genepred[junc['chr2']]
rightend = 'N'
if junc['dir2'] == '+' and genepred_basics.is_exon_start(junc['chr2'],int(junc['coo2']),genepred_entries):
rightend = 'Y'
elif junc['dir2'] == '-' and genepred_basics.is_exon_end(junc['chr2'],int(junc['coo2']),genepred_entries):
rightend = 'Y'
annot1 = genepred_basics.gene_annotate_by_coordinates(junc['chr1'],int(junc['coo1'])-1,int(junc['coo1']),annot_struct)
annot2 = genepred_basics.gene_annotate_by_coordinates(junc['chr2'],int(junc['coo2'])-1,int(junc['coo2']),annot_struct)
best1 = junc['chr1']+":"+str(junc['coo1'])
best1dir = ''
best2 = junc['chr2']+":"+str(junc['coo2'])
best2dir = ''
name1 = ''
name2 = ''
if annot1:
best1 = annot1[0][0]
name1 = annot1[0][0]
best1dir = annot1[0][1]
if annot2:
best2 = annot2[0][0]
name2 = annot2[0][0]
best2dir = annot2[0][1]
evidence_transcript = 'N'
evidence_splice = 'N'
action = 'unknown'
if best1dir == '+' and best2dir == '+':
if junc['dir1'] == '-' and junc['dir2'] == '-':
action= "reverse"
elif junc['dir1'] == '+' and junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
elif best1dir == '-' and best2dir == '-':
if junc['dir1'] == '+' and junc['dir2'] == '+':
action = "reverse"
elif junc['dir1'] == '-' and junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '+' and best2dir == '-':
if junc['dir1'] == '-' and junc['dir2'] == '+':
action = "reverse"
elif junc['dir1'] == '+' and junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '-' and best2dir == '+':
if junc['dir1'] == '+' and junc['dir2'] == '-':
action = "reverse"
elif junc['dir1'] == '-' and junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '-' and best2dir == '':
if junc['dir1'] == '+':
action = "reverse"
elif junc['dir1'] == '-':
action = "no reverse"
else:
action = "unknown"
if best1dir == '+' and best2dir == '':
if junc['dir1'] == '-':
action = "reverse"
elif junc['dir1'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '' and best2dir == '+':
if junc['dir2'] == '-':
action = "reverse"
elif junc['dir2'] == '+':
action = "no reverse"
else:
action = "unknown"
if best1dir == '' and best2dir == '-':
if junc['dir2'] == '+':
action = "reverse"
elif junc['dir2'] == '-':
action = "no reverse"
else:
action = "unknown"
#set if we have transcript evidence
if action != "unknown":
evidence_transcript = 'Y'
# set if we have splice evidence
if action == "no reverse" and splice == 'GT AG':
evidence_splice = 'Y'
if action == "reverse" and splice == 'CT AC':
evidence_splice = 'Y'
if action == "unknown" and (splice == 'GT AG' or splice == 'CT AC'):
evidence_splice = 'Y'
# if action is still unknown, use splice site to make choice to reverse
if action == "unknown":
if splice == 'GT AG':
action = "no reverse"
elif splice == 'CT AC':
action = "reverse"
# switch our left and right end assignments if reverse
if action == "reverse":
temp1 = leftend
leftend = rightend
rightend = temp1
names = ''
if action == "no reverse":
names = junc['chr1'] + "\t" + junc['coo1'] + "\t" + junc['dir1'] + "\t" + leftend + "\t" + junc['chr2'] + "\t" + junc['coo2'] + "\t" + junc['dir2'] + "\t" + rightend + "\t" + best1 + "-" + best2 + "\t" + best1dir + "/" + best2dir + "\t"+evidence_transcript+"\t"+evidence_splice+"\t"+name1 + "\t" + name2
elif action == "reverse":
names = junc['chr2'] + "\t" + junc['coo2'] + "\t" + opposite(junc['dir2']) + "\t" + leftend + "\t" + junc['chr1'] + "\t" + junc['coo1'] + "\t" + opposite(junc['dir1']) + "\t" + rightend+ "\t" +best2 + "-" + best1 + "\t" + best2dir + "/" + best1dir + "\t"+evidence_transcript+"\t"+evidence_splice+"\t" + name2 + "\t" + name1
else:
names = junc['chr1'] + "\t" + junc['coo1'] + "\t" + junc['dir1'] + "\t" + leftend + "\t" +junc['chr2'] + "\t" + junc['coo2'] + "\t" + junc['dir2'] + "\t" + rightend + "\t" + best1 + "-" + best2 + "\t" + best1dir + "/" + best2dir + "\t"+evidence_transcript+"\t"+evidence_splice+"\t" + name1 + "\t" + name2
result[junc_name] = junc_name + "\t" + names + "\t" + str(junc['lr_count']) + "\t" + junc['type1nr'] + "\t" + junc['type1'] + "\t" + junc['type2'] + "\t" + junc['type3'] + "\t" + junc['type4'] + "\t" + str(junc['max_min_side_lengths']) + "\t" + str(junc['max_min_unique_side_counts']) + "\t" + junc['max_min_flanking_distance']
return result