def extract_feature_from_seq(seq,stt,stp,c_tab,g_tab):
'''extract features of sequence from fasta entry'''
stt_coden = stt.strip().split(',')
stp_coden = stp.strip().split(',')
transtab = maketrans("ACGTNX","TGCANX")
mRNA_seq = seq.upper()
mRNA_size = len(seq)
tmp = orf.ORFFinder(mRNA_seq)
(CDS_size1, CDS_frame1, CDS_seq1) = tmp.longest_orf(direction="+",start_coden=stt_coden, stop_coden=stp_coden)
fickett_score1 = fickett.fickett_value(CDS_seq1)
hexamer = FrameKmer.kmer_ratio(CDS_seq1,6,3,c_tab,g_tab)
return (mRNA_size, CDS_size1, fickett_score1,hexamer)
def extract_feature_from_bed(inbed, refgenome, stt, stp, c_tab, g_tab):
'''extract features of sequence from bed line'''
stt_coden = stt.strip().split(',')
stp_coden = stp.strip().split(',')
transtab = str.maketrans("ACGTNX", "TGCANX")
mRNA_seq = ''
mRNA_size = 0
if inbed.strip():
try:
fields = inbed.split()
chrom = fields[0]
tx_start = int(fields[1])
tx_end = int(fields[2])
geneName = fields[3]
strand = fields[5].replace(" ", "_")
exon_num = int(fields[9])
exon_sizes = list(map(int, fields[10].rstrip(',\n').split(',')))
exon_starts = list(map(int, fields[11].rstrip(',\n').split(',')))
exon_starts = list(map((lambda x: x + tx_start), exon_starts))
exon_ends = list(map(int, fields[10].rstrip(',\n').split(',')))
exon_ends = list(map((lambda x, y: x + y), exon_starts, exon_ends))
intron_starts = exon_ends[:-1]
intron_ends = exon_starts[1:]
except:
print("Wrong format!" + inbed, file=sys.stderr)
return None
mRNA_size = sum(exon_sizes)
for st, end in zip(exon_starts, exon_ends):
exon_coord = chrom + ':' + str(st + 1) + '-' + str(end)
tmp = pysam.faidx(refgenome, exon_coord)
mRNA_seq += ''.join([i.rstrip('\n\r') for i in tmp[1:]])
if strand == '-':
mRNA_seq = mRNA_seq.upper().translate(transtab)[::-1]
tmp = orf.ORFFinder(mRNA_seq)
(CDS_size, CDS_frame, CDS_seq) = tmp.longest_orf(direction="+",
start_coden=stt_coden,
stop_coden=stp_coden)
fickett_score = fickett.fickett_value(CDS_seq)
hexamer = FrameKmer.kmer_ratio(CDS_seq, 6, 3, c_tab, g_tab)
#print CDS_seq
return (geneName, mRNA_size, CDS_size, fickett_score, hexamer)
def mainProcess(input, output, number, c_tab, g_tab, codonArr, hash_matrix,
classifier):
if number > 1:
Temp_Dir = output + '_Tmp_Dir'
temp_score = '' + Temp_Dir + '/' + output + str(number)
# temp_feature = ''+Temp_Dir+'/temp_feature' + str(number)
SCORE = open(temp_score, 'w')
# DATA = open(temp_feature,'w')
sequence_Arr = input.split('\n')
sLen = len(sequence_Arr) - 1
del sequence_Arr[sLen]
if number == 1:
SCORE = open(output, 'w')
sequence_Arr = input
label_Arr_tmp = []
FastA_seq_Arr_tmp = []
for n in range(len(sequence_Arr)):
if n == 0 or n % 2 == 0:
label = sequence_Arr[n]
label_Arr_tmp.append(label)
else:
seq = sequence_Arr[n]
FastA_seq_Arr_tmp.append(seq)
data = []
ids = []
for i in range(len(label_Arr_tmp)):
Seq = FastA_seq_Arr_tmp[i]
tran_fir_seq = Seq.lower()
tran_sec_seq_one = tran_fir_seq.replace('u', 't')
strinfo = re.compile('[^agctn]')
tran_sec_seq = strinfo.sub('n', tran_sec_seq_one)
tran_sec_seq2 = tran_sec_seq.upper()
tmp = orf.ORFFinder(tran_sec_seq2)
(CDS_start, CDS_stop, CDS_size, CDS_frame,
CDS_seq) = tmp.longest_orf(direction="+")
(MCS, CSL, CP) = mcssProcess(tran_sec_seq2, c_tab, g_tab)
fickett_score = fickett.fickett_value(CDS_seq)
(orfscore, orfdistance) = HexamerFeatures(CDS_seq.lower(), hash_matrix)
labels_Arr = label_Arr_tmp[i].split()
ids.append(labels_Arr[0])
Exons_mscore = []
Exons_distance = []
Exons_GC = []
Site_start = 0
for j in range(1, len(labels_Arr)):
seq = tran_sec_seq[Site_start:Site_start + int(labels_Arr[j])]
if (len(seq) > 0):
GCnum = seq.count('c') + seq.count('g')
GCratio = GCnum / float(len(seq))
Exons_GC.append(GCratio)
(mscore, distance) = HexamerFeatures(seq, hash_matrix)
Exons_mscore.append(mscore)
Exons_distance.append(distance)
Site_start = Site_start + int(labels_Arr[j])
else:
continue
Max_Mscore_exon = max(Exons_mscore)
Max_distance = max(Exons_distance)
Max_GCcontent = max(Exons_GC)
full_len = len(tran_sec_seq)
orf_ratio = CDS_size / float(full_len)
transcript_features = [
CDS_size, orf_ratio, fickett_score, orfscore, orfdistance,
Max_Mscore_exon, Max_distance, Max_GCcontent, MCS, CSL, CP
]
data.append(transcript_features)
# PROPERTY_STR = labels_Arr[0] + ' ' + str(CDS_size) + ' '+ str(orf_ratio) + ' ' + str(fickett_score) + ' '+ str(orfscore) + ' '+ str(orfdistance)+' '+ str(Max_Mscore_exon)+ ' ' + str(Max_distance)+ ' ' + str(Max_GCcontent)+ ' ' +str(MCS) +' '+str(CSL)+' '+str(CP)+'\n'
# DATA.write(PROPERTY_STR)
testing_data = np.array(data)
del data
testing_data = testing_data.reshape(len(label_Arr_tmp), 11)
prob = classifier.predict_proba(testing_data)
labels = classifier.predict(testing_data)
PrintResult(ids, labels, prob[:, 1], SCORE)
SCORE.close()
def mainProcess(input,output,number,c_tab,g_tab,codonArr,hash_matrix,mRNA_num):
mRNA_num = mRNA_num/2
if number > 1:
Temp_Dir = output + '_Tmp_Dir'
temp_feature = ''+Temp_Dir+'/'+ output + str(number)
DATA = open(temp_feature,'w')
sequence_Arr = input.split('\n')
sLen = len(sequence_Arr) - 1
del sequence_Arr[sLen]
if number == 1:
DATA = open(output,'w')
sequence_Arr = input
label_Arr_tmp = []
FastA_seq_Arr_tmp = []
for n in range(len(sequence_Arr)):
if n == 0 or n % 2 == 0:
label = sequence_Arr[n]
label_Arr_tmp.append(label)
else :
seq = sequence_Arr[n]
FastA_seq_Arr_tmp.append(seq)
for i in range(len(label_Arr_tmp)):
Seq = FastA_seq_Arr_tmp[i]
tran_fir_seq = Seq.lower()
tran_sec_seq_one = tran_fir_seq.replace('u','t')
strinfo = re.compile('[^agctn]')
tran_sec_seq = strinfo.sub('n',tran_sec_seq_one)
tran_sec_seq2 = tran_sec_seq.upper()
tmp = orf.ORFFinder(tran_sec_seq2)
(CDS_start, CDS_stop, CDS_size, CDS_frame, CDS_seq) = tmp.longest_orf(direction="+")
(MCS,CSL,CP) = mcssProcess(tran_sec_seq2,c_tab,g_tab)
fickett_score = fickett.fickett_value(CDS_seq)
(orfscore,orfdistance) = HexamerFeatures(CDS_seq.lower(),hash_matrix)
labels_Arr = label_Arr_tmp[i].split()
Exons_mscore = []
Exons_distance =[]
Exons_GC = []
Site_start = 0
for j in range(1,len(labels_Arr)):
seq = tran_sec_seq[Site_start:Site_start+int(labels_Arr[j])]
if (len(seq) > 0):
GCnum = seq.count('c') + seq.count('g')
GCratio = GCnum/float(len(seq))
Exons_GC.append(GCratio)
(mscore,distance) = HexamerFeatures(seq,hash_matrix)
Exons_mscore.append(mscore)
Exons_distance.append(distance)
Site_start = Site_start + int(labels_Arr[j])
else:
continue
Max_Mscore_exon = max(Exons_mscore)
Max_distance = max(Exons_distance)
Max_GCcontent = max(Exons_GC)
full_len = len(tran_sec_seq)
orf_ratio = CDS_size/float(full_len)
if i < mRNA_num:
PROPERTY_STR = '+1 '+ str(CDS_size) + ' '+ str(orf_ratio) + ' ' + str(fickett_score) + ' '+ str(orfscore) + ' '+ str(orfdistance)+' '+ str(Max_Mscore_exon)+ ' ' + str(Max_distance)+ ' ' + str(Max_GCcontent)+ ' ' +str(MCS) +' '+str(CSL)+' '+str(CP)+'\n'
else:
PROPERTY_STR = '-1 '+ str(CDS_size) + ' '+ str(orf_ratio) + ' ' + str(fickett_score) + ' '+ str(orfscore) + ' '+ str(orfdistance)+' '+ str(Max_Mscore_exon)+ ' ' + str(Max_distance)+ ' ' + str(Max_GCcontent)+ ' ' +str(MCS) +' '+str(CSL)+' '+str(CP)+'\n'
DATA.write(PROPERTY_STR)
DATA.close()
