9
|
1 /*
|
|
2 *
|
|
3 * Sintax: ./frags2text fragsFILE.frags fastaX fastaY fragsFILE.txt
|
|
4
|
|
5 */
|
|
6
|
|
7 #include <stdio.h>
|
|
8 #include <stdlib.h>
|
|
9 #include <string.h>
|
|
10 #include <inttypes.h>
|
|
11 #include "structs.h"
|
|
12 #include "commonFunctions.h"
|
|
13 #include "comparisonFunctions.h"
|
|
14
|
|
15 #define TAB_INSERT 70
|
|
16 #define READING_FRAG_BUFFER 10000
|
|
17
|
|
18 #define MAX(a,b) \
|
|
19 ({ __typeof__ (a) _a = (a); \
|
|
20 __typeof__ (b) _b = (b); \
|
|
21 _a > _b ? _a : _b; })
|
|
22
|
|
23 #define MIN(a,b) \
|
|
24 ({ __typeof__ (a) _a = (a); \
|
|
25 __typeof__ (b) _b = (b); \
|
|
26 _a < _b ? _a : _b; })
|
|
27
|
|
28 void csv_frag_to_struct_frag(char * l, struct FragFile * f){
|
|
29
|
|
30 //0 1 2 3 4 5 6 7 8 9 10 11 12 13
|
|
31 //Type xStart yStart xEnd yEnd strand(f/r) block length score ident similarity %ident SeqX SeqY
|
|
32 //Frag 3147493 3006054 3154663 2998884 f 0 7171 20868 6194 72.75 0.86 0 0
|
|
33
|
|
34
|
|
35 float bin;
|
|
36
|
|
37 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);
|
|
38
|
|
39 //printf("Read %d items\n", items);
|
|
40
|
|
41 }
|
|
42
|
|
43 struct rIndex2 * loadReadsIndex(char * filename, uint64_t * nReads){
|
|
44 struct rIndex2 * RR;
|
|
45 uint64_t nR=0,i;
|
|
46 FILE *f;
|
|
47 uint64_t fsize;
|
|
48
|
|
49 if ((f=fopen(filename,"rb"))==NULL) terror("Could not open index input file");
|
|
50
|
|
51 fseeko(f, 0, SEEK_END);
|
|
52 fsize = ftello(f);
|
|
53 rewind(f);
|
|
54 nR = fsize/sizeof(struct rIndex2);
|
|
55
|
|
56 if ((RR =(struct rIndex2*) calloc(nR,sizeof(struct rIndex2)))==NULL) terror("Could not allocate index");
|
|
57
|
|
58 for (i=0; i<nR; i++){
|
|
59 if(0 == fread(&RR[i],sizeof(struct rIndex2),1,f)) break;
|
|
60 }
|
|
61 fclose(f);
|
|
62 (*nReads) = nR;
|
|
63 return RR;
|
|
64 }
|
|
65
|
|
66 void write_headers(FILE * f1, FILE * f2, FILE * fO, uint64_t pos1, uint64_t pos2, struct FragFile * f){
|
|
67 fseek(f1, pos1, SEEK_SET);
|
|
68 char c = fgetc(f1);
|
|
69 while(c != '\n'){ fprintf(fO, "%c", c); c = fgetc(f1); }
|
|
70 fprintf(fO, " ALIGNED WITH ");
|
|
71 fseek(f2, pos2, SEEK_SET);
|
|
72 c = fgetc(f2);
|
|
73 while(c != '\n'){ fprintf(fO, "%c", c); c = fgetc(f2); }
|
|
74 fprintf(fO, " LENGTH: %"PRIu64" IDENT: %"PRIu64" STRAND: %c @(%"PRIu64", %"PRIu64")\n", f->length, f->ident, f->strand, f->xStart, f->yStart);
|
|
75 }
|
|
76
|
|
77 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){
|
|
78 char copyX[TAB_INSERT+1];
|
|
79 char copyY[TAB_INSERT+1];
|
|
80 memset(copyX, 0x0, TAB_INSERT+1);
|
|
81 memset(copyY, 0x0, TAB_INSERT+1);
|
|
82 uint64_t atcopyX = 0;
|
|
83 uint64_t atcopyY = 0;
|
|
84 uint64_t i;
|
|
85
|
|
86 uint64_t pos_used_x, pos_used_y;
|
|
87
|
|
88 // The X one
|
|
89 //fseek(fastaX, posX, SEEK_SET);
|
|
90 pos_used_x = posX;
|
|
91 char cX = fastaX[pos_used_x++];
|
|
92 while(cX != '\n'){ cX = fastaX[pos_used_x++]; }
|
|
93 uint64_t gpX = iniX - LacX;
|
|
94 uint64_t currX = 0, tab_counter;
|
|
95 while(currX < gpX){
|
|
96 cX = fastaX[pos_used_x++];
|
|
97 if(cX != '\n') ++currX;
|
|
98 }
|
|
99
|
|
100 // the other Y
|
|
101 //fseek(fastaY, posY, SEEK_SET);
|
|
102 pos_used_y = posY;
|
|
103 char cY = fastaY[pos_used_y++];
|
|
104 while(cY != '\n'){ cY = fastaY[pos_used_y++]; }
|
|
105 uint64_t gpY = iniY - LacY;
|
|
106 uint64_t currY = 0;
|
|
107 while(currY < gpY){
|
|
108 cY = fastaY[pos_used_y++];
|
|
109 if(cY != '\n') ++currY;
|
|
110 }
|
|
111
|
|
112
|
|
113
|
|
114
|
|
115 // Reached the region to show
|
|
116 currX = 0;
|
|
117 currY = 0;
|
|
118 cX = fastaX[pos_used_x++];
|
|
119 cY = fastaY[pos_used_y++];
|
|
120 //fprintf(fO, "\t");
|
|
121 tab_counter = 0;
|
|
122 while(currX < lX && currY < lY){
|
|
123
|
|
124 if(cX == 'A' || cX == 'C' || cX == 'G' || cX == 'T' || cX == 'N'){ copyX[atcopyX++] = cX; ++currX; }
|
|
125 cX = fastaX[pos_used_x++];
|
|
126 while(cX != 'A' && cX != 'C' && cX != 'G' && cX != 'T' && cX != 'N') cX = fastaX[pos_used_x++];
|
|
127
|
|
128
|
|
129 if(cY == 'A' || cY == 'C' || cY == 'G' || cY == 'T' || cY == 'N'){ copyY[atcopyY++] = cY; ++currY; }
|
|
130 cY = fastaY[pos_used_y++];
|
|
131 while(cY != 'A' && cY != 'C' && cY != 'G' && cY != 'T' && cY != 'N') cY = fastaY[pos_used_y++];
|
|
132
|
|
133 while(currX > currY){
|
|
134 if(cY == 'A' || cY == 'C' || cY == 'G' || cY == 'T' || cY == 'N'){ copyY[atcopyY++] = cY; ++currY; }
|
|
135 cY = fastaY[pos_used_y++];
|
|
136 while(cY != 'A' && cY != 'C' && cY != 'G' && cY != 'T' && cY != 'N') cY = fastaY[pos_used_y++];
|
|
137 }
|
|
138 while(currX < currY){
|
|
139 if(cX == 'A' || cX == 'C' || cX == 'G' || cX == 'T' || cX == 'N'){ copyX[atcopyX++] = cX; ++currX; }
|
|
140 cX = fastaX[pos_used_x++];
|
|
141 while(cX != 'A' && cX != 'C' && cX != 'G' && cX != 'T' && cX != 'N') cX = fastaX[pos_used_x++];
|
|
142 }
|
|
143
|
|
144 ++tab_counter;
|
|
145
|
|
146
|
|
147
|
|
148 if(tab_counter >= TAB_INSERT && currX == currY){
|
|
149
|
|
150 copyX[TAB_INSERT] = '\0';
|
|
151 copyY[TAB_INSERT] = '\0';
|
|
152 fprintf(fO, "X:\t%.*s\n\t", TAB_INSERT, copyX);
|
|
153 for(i=0; i<TAB_INSERT; i++){
|
|
154 if(copyX[i] == copyY[i]) fprintf(fO, "|"); else fprintf(fO, " ");
|
|
155 }
|
|
156 fprintf(fO, "\nY:\t%.*s\n\n", TAB_INSERT, copyY);
|
|
157 tab_counter = 0;
|
|
158 atcopyX = 0; atcopyY = 0;
|
|
159 }
|
|
160 }
|
|
161 if(atcopyX > 0){
|
|
162 copyX[atcopyX] = '\0';
|
|
163 copyY[atcopyY] = '\0';
|
|
164 fprintf(fO, "X:\t%.*s\n\t", (int)atcopyX, copyX);
|
|
165 for(i=0; i<atcopyX; i++){
|
|
166 if(copyX[i] == copyY[i]) fprintf(fO, "|"); else fprintf(fO, " ");
|
|
167 }
|
|
168 fprintf(fO, "\nY:\t%.*s\n\n", (int)atcopyY, copyY);
|
|
169 atcopyX = 0; atcopyY = 0;
|
|
170 }
|
|
171 fprintf(fO, "\n");
|
|
172 }
|
|
173
|
|
174
|
|
175 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){
|
|
176 fseek(fasta, pos, SEEK_SET);
|
|
177 char c = fgetc(fasta);
|
|
178 while(c != '\n'){ c = fgetc(fasta); }
|
|
179 uint64_t gp = ini - Lac;
|
|
180 uint64_t curr = 0, tab_counter;
|
|
181 while(curr < gp){
|
|
182 c = fgetc(fasta);
|
|
183 if(c != '\n') ++curr;
|
|
184 }
|
|
185 // Reached the region to show
|
|
186 curr = 0;
|
|
187 c = fgetc(fasta);
|
|
188 fprintf(fO, "\t");
|
|
189 tab_counter = 0;
|
|
190 while(curr < l){
|
|
191 if(c != '\n') fprintf(fO, "%c", c);
|
|
192 c = fgetc(fasta);
|
|
193 if(c != '\n' || feof(fasta)){ ++curr; ++tab_counter; }
|
|
194 if(tab_counter == TAB_INSERT){ fprintf(fO, "\n\t"); tab_counter = 0;}
|
|
195 }
|
|
196 fprintf(fO, "\n");
|
|
197 }
|
|
198
|
|
199
|
|
200 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){
|
|
201 fseek(fasta, pos, SEEK_SET);
|
|
202 char c = fgetc(fasta);
|
|
203 while(c != '\n'){ c = fgetc(fasta); }
|
|
204 uint64_t gp = ini - Lac;
|
|
205 uint64_t curr = 0, tab_counter;
|
|
206 while(curr < gp){
|
|
207 c = fgetc(fasta);
|
|
208 if(c != '\n') ++curr;
|
|
209 }
|
|
210 // Reached the region to show
|
|
211 curr = 0;
|
|
212 c = fgetc(fasta);
|
|
213 fprintf(fO, "\t");
|
|
214 tab_counter = 0;
|
|
215 while(curr < l){
|
|
216 if(c != '\n') fprintf(fO, "%c", c);
|
|
217 c = fgetc(fasta);
|
|
218 if(c != '\n' || feof(fasta)){ ++curr; ++tab_counter; }
|
|
219 if(tab_counter == TAB_INSERT){ fprintf(fO, "\n\t"); tab_counter = 0;}
|
|
220 }
|
|
221 fprintf(fO, "\n");
|
|
222 }
|
|
223
|
|
224
|
|
225
|
|
226 int main(int ac, char** av) {
|
|
227 FILE* fFrags;
|
|
228 struct FragFile frag;
|
|
229
|
|
230 //fprintf(stdout, "[WARNING] - Remember that if using a CSV make sure that the reverse y coordinates are transformed\n");
|
|
231
|
|
232 if (ac != 10)
|
|
233 terror("USE: ./frags2text fragsFILE.csv fastaX fastaY fastaYrev indexX indexY indexYrev fragsFILE.txt borderSize");
|
|
234
|
|
235 // prepared for multiple files
|
|
236 if ((fFrags = fopen(av[1], "rt")) == NULL)
|
|
237 terror("Opening Frags csv file");
|
|
238
|
|
239 // Open fastas
|
|
240
|
|
241 FILE * fX = NULL, * fY = NULL, * fYrev = NULL, * fO = NULL;
|
|
242 fX = fopen(av[2], "rt");
|
|
243 if(fX == NULL) terror("Could not open fasta X file");
|
|
244 fY = fopen(av[3], "rt");
|
|
245 if(fY == NULL) terror("Could not open fasta Y file");
|
|
246 fYrev = fopen(av[4], "rt");
|
|
247 if(fYrev == NULL) terror("Could not open fasta Y-rev file");
|
|
248
|
|
249
|
|
250 // Get file lengths
|
|
251 fseek(fX, 0, SEEK_END);
|
|
252 uint64_t aprox_lenX = ftell(fX);
|
|
253 rewind(fX);
|
|
254 char * strfastaX = (char *) malloc(aprox_lenX*sizeof(char));
|
|
255 fseek(fY, 0, SEEK_END);
|
|
256 uint64_t aprox_lenY = ftell(fY);
|
|
257 rewind(fY);
|
|
258 char * strfastaY = (char *) malloc(aprox_lenY*sizeof(char));
|
|
259 fseek(fYrev, 0, SEEK_END);
|
|
260 uint64_t aprox_lenYrev = ftell(fYrev);
|
|
261 rewind(fYrev);
|
|
262 char * strfastaYrev = (char *) malloc(aprox_lenYrev*sizeof(char));
|
|
263
|
|
264 if(strfastaX == NULL || strfastaY == NULL || strfastaYrev == NULL) terror("Could not allocate string sequences");
|
|
265
|
|
266 if(aprox_lenX != fread(strfastaX, sizeof(char), aprox_lenX, fX)) terror("Read wrong number of chars at X sequence");
|
|
267 if(aprox_lenY != fread(strfastaY, sizeof(char), aprox_lenY, fY)) terror("Read wrong number of chars at Y sequence");
|
|
268 if(aprox_lenYrev != fread(strfastaYrev, sizeof(char), aprox_lenYrev, fYrev)) terror("Read wrong number of chars at Y reversed sequence");
|
|
269
|
|
270
|
|
271 struct rIndex2 * RI_X, * RI_Y, * RI_Yrev;
|
|
272
|
|
273
|
|
274 uint64_t nReads_X, nReads_Y;
|
|
275 RI_X = loadReadsIndex(av[5], &nReads_X);
|
|
276 RI_Y = loadReadsIndex(av[6], &nReads_Y);
|
|
277 RI_Yrev = loadReadsIndex(av[7], &nReads_Y);
|
|
278
|
|
279 fO = fopen(av[8], "wt");
|
|
280 if(fO == NULL) terror("Could not open output alignments file");
|
|
281
|
|
282 uint64_t border_size = (uint64_t) atoi(av[9]);
|
|
283
|
|
284
|
|
285 //readSequenceLength(&n1, fFrags);
|
|
286 //readSequenceLength(&n2, fFrags);
|
|
287 // Skip first lines
|
|
288 char buffer[READING_FRAG_BUFFER];
|
|
289 // Lengths are in lines 6 and 7
|
|
290 //SeqX length : 249250621
|
|
291 int64_t xlen, ylen;
|
|
292 int skipper = 0; while(skipper < 18){
|
|
293 fgets(buffer, READING_FRAG_BUFFER, fFrags);
|
|
294
|
|
295 if(skipper == 6){
|
|
296 sscanf(buffer, "%*s %*s : %"PRId64, &xlen);
|
|
297 }
|
|
298
|
|
299 if(skipper == 7){
|
|
300 sscanf(buffer, "%*s %*s : %"PRId64, &ylen);
|
|
301 }
|
|
302 skipper++;
|
|
303 }
|
|
304
|
|
305 fprintf(stdout, "Sequence lengths: (%"PRId64", %"PRId64")\n", xlen, ylen);
|
|
306
|
|
307
|
|
308 //readFragment(&frag, fFrags);
|
|
309
|
|
310
|
|
311 csv_frag_to_struct_frag(buffer, &frag);
|
|
312
|
|
313 // RI_X is the forward index for fasta file X
|
|
314 // RI_Y is the forward index for fasta file Y
|
|
315 // RI_Yrev is the reverse index for fasta file Y
|
|
316
|
|
317 int exit = 0;
|
|
318
|
|
319 while (!feof(fFrags) && exit == 0) {
|
|
320
|
|
321 //RI[id].pos is position in file of >
|
|
322 //RI[id].Lac is the sum of the reads length prior
|
|
323
|
|
324 if(frag.strand == 'f'){
|
|
325 write_headers(fX, fY, fO, RI_X[frag.seqX].pos, RI_Y[frag.seqY].pos, &frag);
|
|
326 }else{
|
|
327 write_headers(fX, fYrev, fO, RI_X[frag.seqX].pos, RI_Yrev[(nReads_Y - frag.seqY) - 1].pos, &frag);
|
|
328 }
|
|
329
|
|
330
|
|
331 //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);
|
|
332
|
|
333
|
|
334
|
|
335 if(frag.strand == 'f'){
|
|
336
|
|
337 int64_t leftx = MAX(0, (int64_t)frag.xStart - (int64_t)border_size);
|
|
338 int64_t rightx = MIN(xlen, (int64_t)frag.xEnd + (int64_t)border_size);
|
|
339 int64_t lefty = MAX(0, (int64_t)frag.yStart - (int64_t)border_size);
|
|
340 int64_t righty = MIN(ylen, (int64_t)frag.yEnd + (int64_t) border_size);
|
|
341
|
|
342 int64_t border_left_size = MIN(frag.xStart - leftx, frag.yStart - lefty);
|
|
343 int64_t border_right_size = MIN(rightx - frag.xEnd, righty - frag.yEnd);
|
|
344
|
|
345 get_both_seqs(strfastaX, strfastaY, frag.xStart - border_left_size, frag.xEnd + border_right_size, frag.yStart - border_left_size, frag.yEnd + border_right_size, RI_X[frag.seqX].pos, RI_Y[frag.seqY].pos, RI_X[frag.seqX].Lac, RI_Y[frag.seqY].Lac, frag.length + border_left_size + border_right_size, frag.length + border_left_size + border_right_size, fO);
|
|
346 //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);
|
|
347 }else{
|
|
348 uint64_t seqYnew;
|
|
349 seqYnew = (nReads_Y - frag.seqY) - 1;
|
|
350
|
|
351 frag.yStart = ylen - frag.yStart - 1;
|
|
352 frag.yEnd = ylen - frag.yEnd - 1;
|
|
353
|
|
354 int64_t leftx = MAX(0, (int64_t)frag.xStart - (int64_t)border_size);
|
|
355 int64_t rightx = MIN(xlen, (int64_t)frag.xEnd + (int64_t)border_size);
|
|
356 int64_t lefty = MAX(0, (int64_t)frag.yStart - (int64_t)border_size);
|
|
357 int64_t righty = MIN(ylen, (int64_t)frag.yEnd + (int64_t) border_size);
|
|
358
|
|
359 int64_t border_left_size = MIN(frag.xStart - leftx, frag.yStart - lefty);
|
|
360 int64_t border_right_size = MIN(rightx - frag.xEnd, righty - frag.yEnd);
|
|
361
|
|
362
|
|
363
|
|
364 get_both_seqs(strfastaX, strfastaYrev, frag.xStart - border_left_size, frag.xEnd + border_right_size, frag.yStart - border_left_size, frag.yEnd + border_right_size, RI_X[frag.seqX].pos, RI_Yrev[seqYnew].pos, RI_X[frag.seqX].Lac, RI_Yrev[seqYnew].Lac, frag.length + border_left_size + border_right_size, frag.length + border_left_size + border_right_size, fO);
|
|
365 //get_seq_from_to_rev(fYrev, fO, frag.yStart, frag.yEnd, RI_Yrev[seqYnew].pos, RI_Yrev[seqYnew].Lac, seqYnew, frag.length, fO);
|
|
366 }
|
|
367
|
|
368 //readFragment(&frag, fFrags);
|
|
369 if(!feof(fFrags)){
|
|
370 fgets(buffer, READING_FRAG_BUFFER, fFrags);
|
|
371 csv_frag_to_struct_frag(buffer, &frag);
|
|
372
|
|
373
|
|
374 }else{
|
|
375 exit = 1;
|
|
376 }
|
|
377 //printf("line read: %s\n", buffer);
|
|
378 //printf("HI im a frag: %"PRIu64", %"PRIu64" - %"PRIu64", %"PRIu64"\n", frag.xStart, frag.xEnd, frag.yStart, frag.yEnd);
|
|
379 //getchar();
|
|
380 }
|
|
381
|
|
382 fclose(fFrags);
|
|
383 fclose(fX);
|
|
384 fclose(fY);
|
|
385 fclose(fO);
|
|
386
|
|
387 free(RI_X);
|
|
388 free(RI_Y);
|
|
389 free(RI_Yrev);
|
|
390
|
|
391 free(strfastaX);
|
|
392 free(strfastaY);
|
|
393 free(strfastaYrev);
|
|
394
|
|
395 return 0;
|
|
396 }
|