TriplexCLI/fea.py at main · RMB1907/TriplexCLI · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
import numpy as np
import pandas as pd

def construct_kmer():
	ntarr = ("A","C","G","T")

	kmerArray = []


	for n in range(4):
		kmerArray.append(ntarr[n])

	for n in range(4):
		str1 = ntarr[n]
		for m in range(4):
			str2 = str1 + ntarr[m]
			kmerArray.append(str2)
#############################################
	for n in range(4):
		str1 = ntarr[n]
		for m in range(4):
			str2 = str1 + ntarr[m]
			for x in range(4):
				str3 = str2 + ntarr[x]
				kmerArray.append(str3)
#############################################
#change this part for 3mer or 4mer
	for n in range(4):
		str1 = ntarr[n]
		for m in range(4):
			str2 = str1 + ntarr[m]
			for x in range(4):
				str3 = str2 + ntarr[x]
				for y in range(4):
					str4 = str3 + ntarr[y]
					kmerArray.append(str4)
############################################
	for n in range(4):
		str1 = ntarr[n]
		for m in range(4):
			str2 = str1 + ntarr[m]
			for x in range(4):
				str3 = str2 + ntarr[x]
				for y in range(4):
					str4 = str3 + ntarr[y]
					for z in range(4):
						str5 = str4 + ntarr[z]
						kmerArray.append(str5)
####################### 6-mer ##############
	for n in range(4):
		str1 = ntarr[n]
		for m in range(4):
			str2 = str1 + ntarr[m]
			for x in range(4):
				str3 = str2 + ntarr[x]
				for y in range(4):
					str4 = str3 + ntarr[y]
					for z in range(4):
						str5 = str4 + ntarr[z]
						for t in range(4):
							str6 = str5 + ntarr[t]
							kmerArray.append(str6)

	return kmerArray

def kmer_encode(seq,kmerarray):
    result = np.zeros((len(seq),len(kmerarray)))
    for i in range(len(seq)):
        for j in range(len(kmerarray)):
            result[i,j] = seq[i].count(kmerarray[j])/len(seq[i])
    return result

def mer_sin(seq, nc_m, c_m, kmerarray, x):
    l = len(seq) - x + 1
    if l <= 0:
        return 0.0
    log_r = np.zeros((l))
    for i in range(l):
        tempseq = seq[i:i+x]
        if 'N' in tempseq:
            log_r[i] = 0
            continue
        idx = kmerarray.index(tempseq)
        Fc = c_m[int(idx)]
        Fnc = nc_m[int(idx)]
        if Fc == 0 and Fnc == 0:
            log_r[i] = 0
        elif Fc == 0 and Fnc != 0:
            log_r[i] = -1
        elif Fnc == 0 and Fc != 0:
            log_r[i] = 1
        else:
            log_r[i] = np.log(Fc / Fnc)
    miu = sum(log_r) / l
    return miu

def mer_score(seq,nc_m,c_m,kmerarray,x):

    miu = np.zeros((len(seq)))
    for i in range(len(seq)):
        miu[i] = mer_sin(seq[i],nc_m,c_m,kmerarray,x)

    miu0 = np.expand_dims(miu, axis=1)
    return miu0

def generate_features(sequences, seq_type):

    if seq_type=='RNA':
        pos = pd.read_csv('embed/mer_rnapos_mean.csv', header=None).values
        neg = pd.read_csv('embed/mer_rnaneg_mean.csv', header=None).values

    if seq_type=='DNA':
        pos = pd.read_csv('embed/mer_dnapos_mean.csv', header=None).values
        neg = pd.read_csv('embed/mer_dnaneg_mean.csv', header=None).values

    kmerArray = construct_kmer()

    # ---- k-mer encodings ----
    kmer1 = kmer_encode(sequences, kmerArray[0:4])      # 4 features
    kmer2 = kmer_encode(sequences, kmerArray[4:20])     # 16 features
    kmer3 = kmer_encode(sequences, kmerArray[20:84])    # 64 features

    # ---- mer scores ----
    merscore1 = mer_score(sequences, pos[0:4], neg[0:4], kmerArray[0:4], 1)
    print('merscore1')
    merscore2 = mer_score(sequences, pos[4:20], neg[4:20], kmerArray[4:20], 2)
    print('merscore2')
    merscore3 = mer_score(sequences, pos[20:84], neg[20:84], kmerArray[20:84], 3)
    print('merscore3')
    merscore4 = mer_score(sequences, pos[84:340], neg[84:340], kmerArray[84:340], 4)
    print('merscore4')
    merscore5 = mer_score(sequences, pos[340:1364], neg[340:1364], kmerArray[340:1364], 5)
    print('merscore5')
    merscore6 = mer_score(sequences, pos[1364:5460], neg[1364:5460], kmerArray[1364:5460], 6)
    print('merscore6')

    feat = np.concatenate(
            (
                merscore1, merscore2, merscore3,
                merscore4, merscore5, merscore6,
                kmer1, kmer2, kmer3
            ),
            axis=1
        )

    input_vector = np.expand_dims(feat, axis=2).astype(np.float32)

    return input_vector