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
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
|
#include "soap_snp.h"
Prob_matrix::Prob_matrix(){
int i;
p_matrix = new rate_t [256*256*4*4]; // 8bit: q_max, 8bit: read_len, 4bit: number of types of all mismatch/match 4x4
p_prior = new rate_t [8*4*4]; // 8(ref ACTGNNNN) * diploid(4x4)
base_freq = new rate_t [4]; // 4 base
type_likely = new rate_t [16+1]; //The 17th element rate_t[16] will be used in comparison
type_prob = new rate_t [16+1];
p_rank = new rate_t [64*64*2048]; // 6bit: N; 5bit: n1; 11bit; T1
p_binom = new rate_t [256*256]; // Total * case
for(i=0;i!=256*256*4*4;i++) {
p_matrix[i] = 1.0;
}
for(i=0;i!=8*4*4;i++) {
p_prior[i] = 1.0;
}
for(i=0;i!=4;i++) {
base_freq[i] = 1.0;
}
for(i=0;i!=16+1;i++) {
type_likely[i] = 0.0; // LOG10 Scale
type_prob[i] = 0.0; // LOG10 Scale
}
for(i=0;i!=64*64*2048;i++) {
p_rank[i] = 1.0;
}
for(i=0;i!=256*256;i++) {
p_binom[i] = 1.0;
}
}
Prob_matrix::~Prob_matrix(){
delete [] p_matrix; // 8bit: q_max, 8bit: read_len, 4bit: number of types of all mismatch/match 4x4
delete [] p_prior; // 8(ref ACTGNNNN) * diploid(4x4)
delete [] base_freq; // 4 base
delete [] type_likely; //The 17th element rate_t[16] will be used in comparison
delete [] type_prob;
delete [] p_rank; // 6bit: N; 5bit: n1; 11bit; T1
delete [] p_binom; // Total * case;
}
int Prob_matrix::matrix_gen(std::ifstream & alignment, Parameter * para, Genome * genome) {
// Read Alignment files
Soap_format soap;
ubit64_t * count_matrix = new ubit64_t [256*256*4*4];
map<Chr_name, Chr_info*>::iterator current_chr;
current_chr = genome->chromosomes.end();
ubit64_t ref(0);
std::string::size_type coord;
for(std::string line; getline(alignment, line);) {
std::istringstream s(line);
if( s>> soap ) {
if(soap.get_pos() < 0) {
continue;
}
//cerr<<soap<<endl;
// In the overloaded "+" above, soap.position will be substracted by 1 so that coordiates start from 0
if (current_chr == genome->chromosomes.end() || current_chr->first != soap.get_chr_name()) {
current_chr = genome->chromosomes.find(soap.get_chr_name());
if(current_chr == genome->chromosomes.end()) {
for(map<Chr_name, Chr_info*>::iterator test = genome->chromosomes.begin();test != genome->chromosomes.end();test++) {
cerr<<'!'<<(test->first)<<'!'<<endl;
}
cerr<<"Assertion Failed: Chromosome: !"<<soap.get_chr_name()<<"! NOT found"<<endl;
exit(255);
}
}
else {
;
}
if (soap.get_pos()+soap.get_read_len()>=current_chr->second->length()) {
continue;
}
if (soap.is_unique()) {
for(coord = 0; coord != soap.get_read_len(); coord++) {
if (soap.is_N(coord)) {
;
}
else {
if(! (soap.get_pos()+coord<current_chr->second->length())) {
cerr<<soap<<endl;
cerr<<"The program found the above read has exceed the reference length:\n";
cerr<<"The read is aligned to postion: "<<soap.get_pos()<<" with read length: "<<soap.get_read_len()<<endl;
cerr<<"Reference: "<<current_chr->first<<" FASTA Length: "<<current_chr->second->length()<<endl;
exit(255);
}
ref = current_chr->second->get_bin_base(soap.get_pos()+coord);
if ( (ref&12) !=0 ) {
// This is an N on reference or a dbSNP which should be excluded from calibration
;
}
else {
if(soap.is_fwd()) {
// forward strand
count_matrix[(((ubit64_t)soap.get_qual(coord))<<12) | (coord<<4) | ((ref&0x3)<<2) | (soap.get_base(coord)>>1)&3] += 1;
}
else {
// reverse strand
count_matrix[(((ubit64_t)soap.get_qual(coord))<<12) | ((soap.get_read_len()-1-coord)<<4) | ((ref&0x3)<<2) | (soap.get_base(coord)>>1)&3] += 1;
}
}
}
}
}
}
}
ubit64_t o_base/*o_based base*/, t_base/*theorecical(supposed) base*/, type, sum[4], same_qual_count_by_type[16], same_qual_count_by_t_base[4], same_qual_count_total, same_qual_count_mismatch;
char q_char/*fastq quality char*/;
const ubit64_t sta_pow=10; // minimum number to say statistically powerful
for(q_char=para->q_min; q_char<=para->q_max ;q_char++) {
memset(same_qual_count_by_type, 0, sizeof(ubit64_t)*16);
memset(same_qual_count_by_t_base, 0, sizeof(ubit64_t)*4);
same_qual_count_total = 0;
same_qual_count_mismatch = 0;
for(coord=0; coord != para->read_length ; coord++) {
for(type=0;type!=16;type++) {
// If the sample is small, then we will not consider the effect of read cycle.
same_qual_count_by_type[type] += count_matrix[ ((ubit64_t)q_char<<12) | coord <<4 | type];
same_qual_count_by_t_base[(type>>2)&3] += count_matrix[ ((ubit64_t)q_char<<12) | coord <<4 | type];
same_qual_count_total += count_matrix[ ((ubit64_t)q_char<<12) | coord <<4 | type];
//cerr<<(int)type<<'\t'<<same_qual_count_by_type[type]<<'\t'<<same_qual_count_by_t_base[0]<<endl;
if(type % 5 != 0) {
// Mismatches
same_qual_count_mismatch += count_matrix[ ((ubit64_t)q_char<<12) | coord <<4 | type];
}
}
}
for(coord=0; coord != para->read_length ; coord++) {
//cerr<<(q_char)<<'\t'<<coord;
memset(sum, (ubit64_t)0, sizeof(ubit64_t)*4);
// Count of all ref base at certain coord and quality
for(type=0;type!=16;type++) {
sum[(type>>2)&3] += count_matrix[ ((ubit64_t)q_char<<12) | (coord <<4) | type]; // (type>>2)&3: the ref base
//cerr<<sum[type&3]<<endl;
}
for(t_base=0; t_base!=4; t_base++) {
for(o_base=0; o_base!=4; o_base++) {
//cerr<<'\t'<<count_matrix[ ((ubit64_t)q_char<<12) | (coord <<4) | (t_base<<2) | o_base];
if (count_matrix[ ((ubit64_t)q_char<<12) | (coord <<4) | (t_base<<2) | o_base] > sta_pow) {
// Statistically powerful
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = ((double)count_matrix[ ((ubit64_t)q_char<<12) | (coord <<4) | (t_base<<2) | o_base]) / sum[t_base];
}
else if (same_qual_count_by_type[t_base<<2|o_base] > sta_pow) {
// Smaller sample, given up effect from read cycle
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = ((double)same_qual_count_by_type[t_base<<2|o_base]) / same_qual_count_by_t_base[t_base];
}
else if (same_qual_count_total > 0){
// Too small sample, given up effect of mismatch types
if (o_base == t_base) {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = ((double)(same_qual_count_total-same_qual_count_mismatch))/same_qual_count_total;
}
else {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = ((double)same_qual_count_mismatch)/same_qual_count_total;
}
}
else {
;
}
// For these cases like:
// Ref: G o_base: G x10 Ax5. When calculate the probability of this allele to be A,
// If there's no A in reference gives observation of G, then the probability will be zero,
// And therefore exclude the possibility of this pos to have an A
// These cases should be avoid when the dataset is large enough
// If no base with certain quality is o_based, it also doesn't matter
if( (p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base]==0) || p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] ==1) {
if (o_base == t_base) {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = (1-pow(10, -((q_char-para->q_min)/10.0)));
if(p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base]<0.25) {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = 0.25;
}
}
else {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = (pow(10, -((q_char-para->q_min)/10.0))/3);
if(p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base]>0.25) {
p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | (t_base<<2) | o_base] = 0.25;
}
}
}
}
}
}
//cerr<<'\t'<<same_qual_count_by_type[0]<<'\t'<<same_qual_count_by_t_base[0]<<endl;
}
delete [] count_matrix;
//memset(&p_matrix[((ubit64_t)para->q_max-para->q_min+1)<<12], 0, 256*256*4*4 - (para->q_max-para->q_min+1)*256*4*4);
// Note: from now on, the first 8 bit of p_matrix is its quality score, not the FASTQ char
return 1;
}
int Prob_matrix::matrix_read(std::fstream &mat_in, Parameter * para) {
int q_char, type;
std::string::size_type coord;
for(std::string line; getline(mat_in, line);) {
std::istringstream s(line);
s>>q_char>>coord;
for(type=0;type!=16;type++) {
s>>p_matrix [ ((ubit64_t)q_char<<12) | (coord <<4) | type];
//cerr<<q_char<<"|"<<coord<<"|"<<p_matrix [ ((ubit64_t)q_char<<12) | (coord <<4) | type]<<endl;
//exit(0);
}
}
return 1;
}
int Prob_matrix::matrix_write(std::fstream &mat_out, Parameter * para) {
for( char q_char = para->q_min; q_char <= para->q_max; q_char++ ) {
for( std::string::size_type coord=0; coord != para->read_length; coord++) {
mat_out<<((ubit64_t)q_char-para->q_min)<<'\t'<<coord;
for(char type=0;type!=16;type++) {
mat_out<<'\t'<<scientific<<showpoint<<setprecision(16)<<p_matrix [ ((ubit64_t)(q_char-para->q_min)<<12) | (coord <<4) | type];
}
mat_out<<endl;
}
}
return 1;
}
|
