def load_manual_threshold_somatics(jsm_file_name, threshold):
'''
Load a list of rows containing somatics based on pre-specified probability threshold.
'''
reader = JointSnvMixReader(jsm_file_name)
chr_list = reader.get_chr_list()
rows = []
scores = []
for chr_name in sorted(chr_list):
print chr_name
chr_rows = reader.get_rows(chr_name)
for row in chr_rows:
score = row['p_aa_ab'] + row['p_aa_bb']
if score >= threshold:
row = format_rows(row, chr_name)
insert_position = bisect.bisect(scores, score)
scores.insert(insert_position, score)
rows.insert(insert_position, row)
reader.close()
return rows
def load_auto_threshold_somatics(jsm_file_name):
'''
Load a list of rows containing somatics based on automatically determined probability threshold. Threshold is
determined based on inflection point method.
'''
n = int(1e5)
threshold = 1e-6
reader = JointSnvMixReader(jsm_file_name)
chr_list = reader.get_chr_list()
scores = []
rows = []
for chr_name in sorted(chr_list):
print chr_name
chr_rows = reader.get_rows(chr_name)
for row in chr_rows:
score = row['p_aa_ab'] + row['p_aa_bb']
insert_position = bisect.bisect(scores, score)
if insert_position > 0 or len(scores) == 0:
scores.insert(insert_position, score)
row = format_rows(row, chr_name)
rows.insert(insert_position, row)
if scores[0] <= threshold or len(scores) > n:
scores.pop(0)
rows.pop(0)
reader.close()
max_diff = 0
index = 0
for i in range(len(scores) - 1):
diff = scores[i + 1] - scores[i]
if diff > max_diff:
max_diff = diff
index = i
rows = rows[index:]
return rows
def load_somatics( jsm_file_name ):
n = int( 1e5 )
threshold = 1e-6
reader = JointSnvMixReader( jsm_file_name )
chr_list = reader.get_chr_list()
scores = []
for chr_name in sorted( chr_list ):
if chr_name in excluded_chrom:
continue
print chr_name
chr_rows = reader.get_rows( chr_name )
for row in chr_rows:
score = row['p_aa_ab'] + row['p_aa_bb']
insert_position = bisect.bisect( scores, score )
if insert_position > 0 or len( scores ) == 0:
scores.insert( insert_position, score )
if scores[0] <= threshold or len( scores ) > n:
scores.pop( 0 )
reader.close()
max_diff = 0
index = 0
for i in range( len( scores ) - 1 ):
diff = scores[i + 1] - scores[i]
if diff > max_diff:
max_diff = diff
index = i
scores = scores[index:]
return scores
