annotate gecko/src/csvFrags2text.c @ 12:7249f6c19612 draft default tip

Uploaded
author bitlab
date Wed, 18 Nov 2020 08:49:53 +0000
parents cf4c0c822ca9
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
9
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
1 /*
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
2 *
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
3 * Sintax: ./frags2text fragsFILE.frags fastaX fastaY fragsFILE.txt
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
4
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
5 */
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
6
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
7 #include <stdio.h>
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
8 #include <stdlib.h>
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
9 #include <string.h>
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
10 #include <inttypes.h>
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
11 #include "structs.h"
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
12 #include "commonFunctions.h"
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
13 #include "comparisonFunctions.h"
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
14
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
15 #define TAB_INSERT 70
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
16 #define READING_FRAG_BUFFER 10000
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
17
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
18 void csv_frag_to_struct_frag(char * l, struct FragFile * f){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
19
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
20 //0 1 2 3 4 5 6 7 8 9 10 11 12 13
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
21 //Type xStart yStart xEnd yEnd strand(f/r) block length score ident similarity %ident SeqX SeqY
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
22 //Frag 3147493 3006054 3154663 2998884 f 0 7171 20868 6194 72.75 0.86 0 0
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
23
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
24
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
25 float bin;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
26
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
27 sscanf(l, "%*s %"PRIu64" %"PRIu64" %"PRIu64" %"PRIu64" %c %"PRId64" %"PRIu64" %"PRIu64" %"PRIu64" %f %f %"PRIu64" %"PRIu64, &f->xStart, &f->yStart, &f->xEnd, &f->yEnd, &f->strand, &f->block, &f->length, &f->score, &f->ident, &f->similarity, &bin, &f->seqX, &f->seqY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
28
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
29 //printf("Read %d items\n", items);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
30
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
31 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
32
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
33 struct rIndex2 * loadReadsIndex(char * filename, uint64_t * nReads){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
34 struct rIndex2 * RR;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
35 uint64_t nR=0,i;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
36 FILE *f;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
37 uint64_t fsize;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
38
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
39 if ((f=fopen(filename,"rb"))==NULL) terror("Could not open index input file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
40
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
41 fseeko(f, 0, SEEK_END);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
42 fsize = ftello(f);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
43 rewind(f);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
44 nR = fsize/sizeof(struct rIndex2);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
45
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
46 if ((RR =(struct rIndex2*) calloc(nR,sizeof(struct rIndex2)))==NULL) terror("Could not allocate index");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
47
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
48 for (i=0; i<nR; i++){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
49 if(0 == fread(&RR[i],sizeof(struct rIndex2),1,f)) break;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
50 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
51 fclose(f);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
52 (*nReads) = nR;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
53 return RR;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
54 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
55
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
56 void write_headers(FILE * f1, FILE * f2, FILE * fO, uint64_t pos1, uint64_t pos2, struct FragFile * f){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
57 fseek(f1, pos1, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
58 char c = fgetc(f1);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
59 while(c != '\n'){ fprintf(fO, "%c", c); c = fgetc(f1); }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
60 fprintf(fO, " ALIGNED WITH ");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
61 fseek(f2, pos2, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
62 c = fgetc(f2);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
63 while(c != '\n'){ fprintf(fO, "%c", c); c = fgetc(f2); }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
64 fprintf(fO, " LENGTH: %"PRIu64" IDENT: %"PRIu64" STRAND: %c @(%"PRIu64", %"PRIu64")\n", f->length, f->ident, f->strand, f->xStart, f->yStart);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
65 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
66
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
67 void get_both_seqs(char * fastaX, char * fastaY, uint64_t iniX, uint64_t finX, uint64_t iniY, uint64_t finY, uint64_t posX, uint64_t posY, uint64_t LacX, uint64_t LacY, uint64_t lX, uint64_t lY, FILE * fO){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
68 char copyX[TAB_INSERT+1];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
69 char copyY[TAB_INSERT+1];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
70 memset(copyX, 0x0, TAB_INSERT+1);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
71 memset(copyY, 0x0, TAB_INSERT+1);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
72 uint64_t atcopyX = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
73 uint64_t atcopyY = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
74 uint64_t i;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
75
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
76 uint64_t pos_used_x, pos_used_y;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
77
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
78 // The X one
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
79 //fseek(fastaX, posX, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
80 pos_used_x = posX;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
81 char cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
82 while(cX != '\n'){ cX = fastaX[pos_used_x++]; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
83 uint64_t gpX = iniX - LacX;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
84 uint64_t currX = 0, tab_counter;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
85 while(currX < gpX){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
86 cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
87 if(cX != '\n') ++currX;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
88 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
89
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
90 // the other Y
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
91 //fseek(fastaY, posY, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
92 pos_used_y = posY;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
93 char cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
94 while(cY != '\n'){ cY = fastaY[pos_used_y++]; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
95 uint64_t gpY = iniY - LacY;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
96 uint64_t currY = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
97 while(currY < gpY){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
98 cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
99 if(cY != '\n') ++currY;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
100 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
101
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
102
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
103
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
104
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
105 // Reached the region to show
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
106 currX = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
107 currY = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
108 cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
109 cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
110 //fprintf(fO, "\t");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
111 tab_counter = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
112 while(currX < lX && currY < lY){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
113
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
114 if(cX == 'A' || cX == 'C' || cX == 'G' || cX == 'T' || cX == 'N'){ copyX[atcopyX++] = cX; ++currX; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
115 cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
116 while(cX != 'A' && cX != 'C' && cX != 'G' && cX != 'T' && cX != 'N') cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
117
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
118
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
119 if(cY == 'A' || cY == 'C' || cY == 'G' || cY == 'T' || cY == 'N'){ copyY[atcopyY++] = cY; ++currY; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
120 cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
121 while(cY != 'A' && cY != 'C' && cY != 'G' && cY != 'T' && cY != 'N') cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
122
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
123 while(currX > currY){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
124 if(cY == 'A' || cY == 'C' || cY == 'G' || cY == 'T' || cY == 'N'){ copyY[atcopyY++] = cY; ++currY; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
125 cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
126 while(cY != 'A' && cY != 'C' && cY != 'G' && cY != 'T' && cY != 'N') cY = fastaY[pos_used_y++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
127 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
128 while(currX < currY){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
129 if(cX == 'A' || cX == 'C' || cX == 'G' || cX == 'T' || cX == 'N'){ copyX[atcopyX++] = cX; ++currX; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
130 cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
131 while(cX != 'A' && cX != 'C' && cX != 'G' && cX != 'T' && cX != 'N') cX = fastaX[pos_used_x++];
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
132 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
133
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
134 ++tab_counter;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
135
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
136
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
137
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
138 if(tab_counter >= TAB_INSERT && currX == currY){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
139
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
140 copyX[TAB_INSERT] = '\0';
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
141 copyY[TAB_INSERT] = '\0';
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
142 fprintf(fO, "X:\t%.*s\n\t", TAB_INSERT, copyX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
143 for(i=0; i<TAB_INSERT; i++){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
144 if(copyX[i] == copyY[i]) fprintf(fO, "|"); else fprintf(fO, " ");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
145 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
146 fprintf(fO, "\nY:\t%.*s\n\n", TAB_INSERT, copyY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
147 tab_counter = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
148 atcopyX = 0; atcopyY = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
149 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
150 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
151 if(atcopyX > 0){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
152 copyX[atcopyX] = '\0';
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
153 copyY[atcopyY] = '\0';
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
154 fprintf(fO, "X:\t%.*s\n\t", (int)atcopyX, copyX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
155 for(i=0; i<atcopyX; i++){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
156 if(copyX[i] == copyY[i]) fprintf(fO, "|"); else fprintf(fO, " ");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
157 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
158 fprintf(fO, "\nY:\t%.*s\n\n", (int)atcopyY, copyY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
159 atcopyX = 0; atcopyY = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
160 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
161 fprintf(fO, "\n");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
162 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
163
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
164
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
165 void get_seq_from_to(FILE * fasta, FILE * output, uint64_t ini, uint64_t fin, uint64_t pos, uint64_t Lac, uint64_t seqNum, uint64_t l, FILE * fO){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
166 fseek(fasta, pos, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
167 char c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
168 while(c != '\n'){ c = fgetc(fasta); }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
169 uint64_t gp = ini - Lac;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
170 uint64_t curr = 0, tab_counter;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
171 while(curr < gp){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
172 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
173 if(c != '\n') ++curr;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
174 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
175 // Reached the region to show
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
176 curr = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
177 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
178 fprintf(fO, "\t");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
179 tab_counter = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
180 while(curr < l){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
181 if(c != '\n') fprintf(fO, "%c", c);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
182 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
183 if(c != '\n' || feof(fasta)){ ++curr; ++tab_counter; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
184 if(tab_counter == TAB_INSERT){ fprintf(fO, "\n\t"); tab_counter = 0;}
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
185 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
186 fprintf(fO, "\n");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
187 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
188
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
189
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
190 void get_seq_from_to_rev(FILE * fasta, FILE * output, uint64_t ini, uint64_t fin, uint64_t pos, uint64_t Lac, uint64_t seqNum, uint64_t l, FILE * fO){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
191 fseek(fasta, pos, SEEK_SET);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
192 char c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
193 while(c != '\n'){ c = fgetc(fasta); }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
194 uint64_t gp = ini - Lac;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
195 uint64_t curr = 0, tab_counter;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
196 while(curr < gp){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
197 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
198 if(c != '\n') ++curr;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
199 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
200 // Reached the region to show
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
201 curr = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
202 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
203 fprintf(fO, "\t");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
204 tab_counter = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
205 while(curr < l){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
206 if(c != '\n') fprintf(fO, "%c", c);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
207 c = fgetc(fasta);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
208 if(c != '\n' || feof(fasta)){ ++curr; ++tab_counter; }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
209 if(tab_counter == TAB_INSERT){ fprintf(fO, "\n\t"); tab_counter = 0;}
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
210 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
211 fprintf(fO, "\n");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
212 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
213
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
214
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
215
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
216 int main(int ac, char** av) {
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
217 FILE* fFrags;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
218 struct FragFile frag;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
219
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
220 fprintf(stdout, "[WARNING] - Remember that if using a CSV make sure that the reverse y coordinates are transformed\n");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
221
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
222 if (ac != 9)
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
223 terror("USE: ./frags2text fragsFILE.csv fastaX fastaY fastaYrev indexX indexY indexYrev fragsFILE.txt");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
224
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
225 // prepared for multiple files
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
226 if ((fFrags = fopen(av[1], "rt")) == NULL)
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
227 terror("Opening Frags csv file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
228
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
229 // Open fastas
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
230
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
231 FILE * fX = NULL, * fY = NULL, * fYrev = NULL, * fO = NULL;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
232 fX = fopen(av[2], "rt");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
233 if(fX == NULL) terror("Could not open fasta X file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
234 fY = fopen(av[3], "rt");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
235 if(fY == NULL) terror("Could not open fasta Y file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
236 fYrev = fopen(av[4], "rt");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
237 if(fYrev == NULL) terror("Could not open fasta Y-rev file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
238
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
239
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
240 // Get file lengths
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
241 fseek(fX, 0, SEEK_END);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
242 uint64_t aprox_lenX = ftell(fX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
243 rewind(fX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
244 char * strfastaX = (char *) malloc(aprox_lenX*sizeof(char));
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
245 fseek(fY, 0, SEEK_END);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
246 uint64_t aprox_lenY = ftell(fY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
247 rewind(fY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
248 char * strfastaY = (char *) malloc(aprox_lenY*sizeof(char));
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
249 fseek(fYrev, 0, SEEK_END);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
250 uint64_t aprox_lenYrev = ftell(fYrev);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
251 rewind(fYrev);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
252 char * strfastaYrev = (char *) malloc(aprox_lenYrev*sizeof(char));
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
253
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
254 if(strfastaX == NULL || strfastaY == NULL || strfastaYrev == NULL) terror("Could not allocate string sequences");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
255
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
256 if(aprox_lenX != fread(strfastaX, sizeof(char), aprox_lenX, fX)) terror("Read wrong number of chars at X sequence");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
257 if(aprox_lenY != fread(strfastaY, sizeof(char), aprox_lenY, fY)) terror("Read wrong number of chars at Y sequence");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
258 if(aprox_lenYrev != fread(strfastaYrev, sizeof(char), aprox_lenYrev, fYrev)) terror("Read wrong number of chars at Y reversed sequence");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
259
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
260
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
261 struct rIndex2 * RI_X, * RI_Y, * RI_Yrev;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
262
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
263
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
264 uint64_t nReads_X, nReads_Y;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
265 RI_X = loadReadsIndex(av[5], &nReads_X);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
266 RI_Y = loadReadsIndex(av[6], &nReads_Y);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
267 RI_Yrev = loadReadsIndex(av[7], &nReads_Y);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
268
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
269 fO = fopen(av[8], "wt");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
270 if(fO == NULL) terror("Could not open output alignments file");
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
271
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
272
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
273 //readSequenceLength(&n1, fFrags);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
274 //readSequenceLength(&n2, fFrags);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
275 // Skip first lines
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
276 char buffer[READING_FRAG_BUFFER];
12
7249f6c19612 Uploaded
bitlab
parents: 11
diff changeset
277 int skipper = 0; while(skipper < 18){ if(NULL==fgets(buffer, READING_FRAG_BUFFER, fFrags)) terror("Couldnt read buffer"); skipper++; }
9
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
278
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
279
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
280
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
281
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
282 //readFragment(&frag, fFrags);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
283
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
284
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
285 csv_frag_to_struct_frag(buffer, &frag);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
286
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
287 // RI_X is the forward index for fasta file X
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
288 // RI_Y is the forward index for fasta file Y
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
289 // RI_Yrev is the reverse index for fasta file Y
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
290
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
291 int exit = 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
292
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
293 while (!feof(fFrags) && exit == 0) {
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
294
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
295 //RI[id].pos is position in file of >
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
296 //RI[id].Lac is the sum of the reads length prior
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
297
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
298 if(frag.strand == 'f'){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
299 write_headers(fX, fY, fO, RI_X[frag.seqX].pos, RI_Y[frag.seqY].pos, &frag);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
300 }else{
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
301 write_headers(fX, fYrev, fO, RI_X[frag.seqX].pos, RI_Yrev[(nReads_Y - frag.seqY) - 1].pos, &frag);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
302 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
303
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
304
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
305 //get_seq_from_to(fX, fO, frag.xStart, frag.xEnd, RI_X[frag.seqX].pos, RI_X[frag.seqX].Lac, frag.seqX, frag.length, fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
306
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
307
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
308
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
309 if(frag.strand == 'f'){
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
310 get_both_seqs(strfastaX, strfastaY, frag.xStart, frag.xEnd, frag.yStart, frag.yEnd, RI_X[frag.seqX].pos, RI_Y[frag.seqY].pos, RI_X[frag.seqX].Lac, RI_Y[frag.seqY].Lac, frag.length, frag.length, fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
311 //get_seq_from_to(fY, fO, frag.yStart, frag.yEnd, RI_Y[frag.seqY].pos, RI_Y[frag.seqY].Lac, frag.seqY, frag.length, fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
312 }else{
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
313 uint64_t seqYnew;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
314 seqYnew = (nReads_Y - frag.seqY) - 1;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
315 get_both_seqs(strfastaX, strfastaYrev, frag.xStart, frag.xEnd, frag.yStart, frag.yEnd, RI_X[frag.seqX].pos, RI_Yrev[seqYnew].pos, RI_X[frag.seqX].Lac, RI_Yrev[seqYnew].Lac, frag.length, frag.length, fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
316 //get_seq_from_to_rev(fYrev, fO, frag.yStart, frag.yEnd, RI_Yrev[seqYnew].pos, RI_Yrev[seqYnew].Lac, seqYnew, frag.length, fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
317 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
318
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
319 //readFragment(&frag, fFrags);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
320 if(!feof(fFrags)){
12
7249f6c19612 Uploaded
bitlab
parents: 11
diff changeset
321 if(NULL==fgets(buffer, READING_FRAG_BUFFER, fFrags)) terror("Couldnt read buffer");
9
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
322 csv_frag_to_struct_frag(buffer, &frag);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
323
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
324
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
325 }else{
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
326 exit = 1;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
327 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
328 //printf("line read: %s\n", buffer);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
329 //printf("HI im a frag: %"PRIu64", %"PRIu64" - %"PRIu64", %"PRIu64"\n", frag.xStart, frag.xEnd, frag.yStart, frag.yEnd);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
330 //getchar();
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
331 }
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
332
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
333 fclose(fFrags);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
334 fclose(fX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
335 fclose(fY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
336 fclose(fO);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
337
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
338 free(RI_X);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
339 free(RI_Y);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
340 free(RI_Yrev);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
341
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
342 free(strfastaX);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
343 free(strfastaY);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
344 free(strfastaYrev);
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
345
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
346 return 0;
aec70bb1ae27 Uploaded
bitlab
parents:
diff changeset
347 }