|
0
|
1
|
|
|
2 #include "utils.h"
|
|
|
3
|
|
|
4
|
|
|
5
|
|
|
6 void print_error(char *msg){
|
|
|
7 fprintf(stderr, "\nmrCaNaVaR version %s.\nLast update: %s.\n", VERSION, LAST_UPDATE);
|
|
|
8 fprintf(stderr, "\n%s\n", msg);
|
|
|
9 fprintf(stderr, "Invoke parameter -h for help.\n");
|
|
|
10 exit (0);
|
|
|
11 }
|
|
|
12
|
|
|
13 void set_runmode(enum MODETYPE NEWMODE){
|
|
|
14 if (RUNMODE != NONE && RUNMODE != NEWMODE){
|
|
|
15 print_error("Select one run mode only.\n");
|
|
|
16 }
|
|
|
17 RUNMODE = NEWMODE;
|
|
|
18 }
|
|
|
19
|
|
|
20 void set_gender(enum GENDERTYPE SET){
|
|
|
21 if (GENDER != AUTODETECT && RUNMODE != SET){
|
|
|
22 print_error("Select only one of --xx or --xy.\n");
|
|
|
23 }
|
|
|
24 GENDER= SET;
|
|
|
25 }
|
|
|
26
|
|
|
27 void set_str(char **target, char *source){
|
|
|
28
|
|
|
29 if (*target != NULL) free((*target));
|
|
|
30
|
|
|
31 (*target) = (char *) malloc(sizeof(char) * (strlen(source)+1));
|
|
|
32
|
|
|
33 strncpy((*target), source, (strlen(source)+1));
|
|
|
34
|
|
|
35 }
|
|
|
36
|
|
|
37
|
|
|
38 void init_globals(void){
|
|
|
39 GENOME_FASTA = NULL;
|
|
|
40 GENOME_CONF = NULL;
|
|
|
41 GENOME_GAPS = NULL;
|
|
|
42 RUNMODE = NONE;
|
|
|
43 GENDER = AUTODETECT;
|
|
|
44 MULTGC = 0;
|
|
|
45 MAX_GC_CORR = 20.0;
|
|
|
46 MIN_GC_CORR = 0.05;
|
|
|
47 VERBOSE = 0;
|
|
|
48
|
|
|
49 num_chrom = 0;
|
|
|
50
|
|
|
51 CHECKSAM = 1;
|
|
|
52
|
|
|
53 /* set window size defaults */
|
|
|
54
|
|
|
55 LW_SIZE = 5000;
|
|
|
56 SW_SIZE = 1000;
|
|
|
57 CW_SIZE = 1000;
|
|
|
58 LW_SLIDE = 1000;
|
|
|
59 SW_SLIDE = 1000;
|
|
|
60
|
|
|
61 LW_MEAN = 0.0;
|
|
|
62 SW_MEAN = 0.0;
|
|
|
63 CW_MEAN = 0.0;
|
|
|
64
|
|
|
65 LW_STD = 0.0;
|
|
|
66 SW_STD = 0.0;
|
|
|
67 CW_STD = 0.0;
|
|
|
68
|
|
|
69 CONT_WINDOW = 7;
|
|
|
70 CUT_WINDOW = 6;
|
|
|
71
|
|
|
72 }
|
|
|
73
|
|
|
74
|
|
|
75 FILE *my_fopen(char *fname, char *mode, char GZ){
|
|
|
76 FILE *fp;
|
|
|
77
|
|
|
78
|
|
|
79 if (!GZ)
|
|
|
80 fp = fopen(fname, mode);
|
|
|
81 else
|
|
|
82 fp = gzopen(fname, mode);
|
|
|
83
|
|
|
84 if (fp == NULL){
|
|
|
85 fprintf(stderr, "Unable to open file %s in %s mode.", fname, mode[0]=='w' ? "write" : "read");
|
|
|
86 exit (0);
|
|
|
87 }
|
|
|
88
|
|
|
89 return fp;
|
|
|
90 }
|
|
|
91
|
|
|
92
|
|
|
93 void saveRefConfig(char *configFile){
|
|
|
94
|
|
|
95 int i;
|
|
|
96 int wcnt;
|
|
|
97 char chrom_name_len;
|
|
|
98 int retVal;
|
|
|
99 FILE *config;
|
|
|
100
|
|
|
101 config = my_fopen(configFile, "w", 0);
|
|
|
102
|
|
|
103 /* sort the chromosomes pointer array */
|
|
|
104
|
|
|
105 qsort(chromosomes, num_chrom, sizeof (struct chrom *), compChr);
|
|
|
106
|
|
|
107
|
|
|
108 /* start with the magicNum, I will use this as a format check when loading */
|
|
|
109 retVal = fwrite(&magicNum, sizeof(magicNum), 1, config);
|
|
|
110
|
|
|
111 /* window sizes / slides */
|
|
|
112
|
|
|
113 retVal = fwrite(&LW_SIZE, sizeof(LW_SIZE), 1, config);
|
|
|
114 retVal = fwrite(&SW_SIZE, sizeof(SW_SIZE), 1, config);
|
|
|
115 retVal = fwrite(&CW_SIZE, sizeof(CW_SIZE), 1, config);
|
|
|
116 retVal = fwrite(&LW_SLIDE, sizeof(LW_SLIDE), 1, config);
|
|
|
117 retVal = fwrite(&SW_SLIDE, sizeof(SW_SLIDE), 1, config);
|
|
|
118
|
|
|
119
|
|
|
120 /* reference genome numbers */
|
|
|
121
|
|
|
122 /* number of chromosomes */
|
|
|
123 retVal = fwrite(&num_chrom, sizeof(num_chrom), 1, config);
|
|
|
124
|
|
|
125 /* iterate through chromosomes, write their names, and window counts */
|
|
|
126
|
|
|
127 for (i=0; i<num_chrom; i++){
|
|
|
128 chrom_name_len = (char) strlen(chromosomes[i]->name);
|
|
|
129 retVal = fwrite(&chrom_name_len, sizeof(chrom_name_len), 1, config);
|
|
|
130 retVal = fwrite(chromosomes[i]->name, chrom_name_len * sizeof(char), 1, config);
|
|
|
131
|
|
|
132 retVal = fwrite(&(chromosomes[i]->length), sizeof(chromosomes[i]->length), 1, config);
|
|
|
133 retVal = fwrite(&(chromosomes[i]->lw_cnt), sizeof(chromosomes[i]->lw_cnt), 1, config);
|
|
|
134 retVal = fwrite(&(chromosomes[i]->sw_cnt), sizeof(chromosomes[i]->sw_cnt), 1, config);
|
|
|
135 retVal = fwrite(&(chromosomes[i]->cw_cnt), sizeof(chromosomes[i]->cw_cnt), 1, config);
|
|
|
136
|
|
|
137 /* long windows */
|
|
|
138 for (wcnt=0; wcnt<chromosomes[i]->lw_cnt; wcnt++){
|
|
|
139 retVal = fwrite(&(chromosomes[i]->lw[wcnt].start), sizeof(chromosomes[i]->lw[wcnt].start), 1, config);
|
|
|
140 retVal = fwrite(&(chromosomes[i]->lw[wcnt].end), sizeof(chromosomes[i]->lw[wcnt].end), 1, config);
|
|
|
141 retVal = fwrite(&(chromosomes[i]->lw[wcnt].gc), sizeof(chromosomes[i]->lw[wcnt].gc), 1, config);
|
|
|
142 }
|
|
|
143
|
|
|
144 /* short windows */
|
|
|
145 for (wcnt=0; wcnt<chromosomes[i]->sw_cnt; wcnt++){
|
|
|
146 retVal = fwrite(&(chromosomes[i]->sw[wcnt].start), sizeof(chromosomes[i]->sw[wcnt].start), 1, config);
|
|
|
147 retVal = fwrite(&(chromosomes[i]->sw[wcnt].end), sizeof(chromosomes[i]->sw[wcnt].end), 1, config);
|
|
|
148 retVal = fwrite(&(chromosomes[i]->sw[wcnt].gc), sizeof(chromosomes[i]->sw[wcnt].gc), 1, config);
|
|
|
149 }
|
|
|
150
|
|
|
151 /* copy windows */
|
|
|
152 for (wcnt=0; wcnt<chromosomes[i]->cw_cnt; wcnt++){
|
|
|
153 retVal = fwrite(&(chromosomes[i]->cw[wcnt].start), sizeof(chromosomes[i]->cw[wcnt].start), 1, config);
|
|
|
154 retVal = fwrite(&(chromosomes[i]->cw[wcnt].end), sizeof(chromosomes[i]->cw[wcnt].end), 1, config);
|
|
|
155 retVal = fwrite(&(chromosomes[i]->cw[wcnt].gc), sizeof(chromosomes[i]->cw[wcnt].gc), 1, config);
|
|
|
156 }
|
|
|
157
|
|
|
158 }
|
|
|
159
|
|
|
160 fclose(config);
|
|
|
161
|
|
|
162 }
|
|
|
163
|
|
|
164
|
|
|
165
|
|
|
166
|
|
|
167 void loadRefConfig(char *configFile){
|
|
|
168
|
|
|
169 int i;
|
|
|
170 int wcnt;
|
|
|
171 char chrom_name_len;
|
|
|
172 char readString[MAX_STR];
|
|
|
173 int retVal;
|
|
|
174
|
|
|
175 int isMagicNum;
|
|
|
176
|
|
|
177 FILE *config;
|
|
|
178
|
|
|
179 config = my_fopen(configFile, "r", 0);
|
|
|
180
|
|
|
181 /* start with the magicNum, use this as a format check when loading */
|
|
|
182 retVal = fread(&isMagicNum, sizeof(isMagicNum), 1, config);
|
|
|
183
|
|
|
184 if (isMagicNum != magicNum)
|
|
|
185 print_error("Reference configuration file seems to be invalid or corrupt.\n");
|
|
|
186
|
|
|
187
|
|
|
188 fprintf(stdout, "Loading reference configuration, hold on ... ");
|
|
|
189 fflush(stdout);
|
|
|
190
|
|
|
191 /* window sizes / slides */
|
|
|
192
|
|
|
193 retVal = fread(&LW_SIZE, sizeof(LW_SIZE), 1, config);
|
|
|
194 retVal = fread(&SW_SIZE, sizeof(SW_SIZE), 1, config);
|
|
|
195 retVal = fread(&CW_SIZE, sizeof(CW_SIZE), 1, config);
|
|
|
196 retVal = fread(&LW_SLIDE, sizeof(LW_SLIDE), 1, config);
|
|
|
197 retVal = fread(&SW_SLIDE, sizeof(SW_SLIDE), 1, config);
|
|
|
198
|
|
|
199
|
|
|
200 /* reference genome numbers */
|
|
|
201
|
|
|
202 /* number of chromosomes */
|
|
|
203 retVal = fread(&num_chrom, sizeof(num_chrom), 1, config);
|
|
|
204
|
|
|
205 /* create chromosomes data structure */
|
|
|
206
|
|
|
207 chromosomes = (struct chrom **) malloc(sizeof(struct chrom *) * num_chrom);
|
|
|
208 for (i=0; i<num_chrom; i++)
|
|
|
209 chromosomes[i] = (struct chrom *) malloc(sizeof(struct chrom) * num_chrom);
|
|
|
210
|
|
|
211 /* iterate through chromosomes, write their names, and window counts */
|
|
|
212
|
|
|
213 for (i=0; i<num_chrom; i++){
|
|
|
214
|
|
|
215 retVal = fread(&chrom_name_len, sizeof(chrom_name_len), 1, config);
|
|
|
216
|
|
|
217 retVal = fread(readString, chrom_name_len * sizeof(char), 1, config);
|
|
|
218 readString[chrom_name_len] = 0;
|
|
|
219 trimspace(readString);
|
|
|
220
|
|
|
221 set_str(&(chromosomes[i]->name), readString);
|
|
|
222
|
|
|
223 retVal = fread(&(chromosomes[i]->length), sizeof(chromosomes[i]->length), 1, config);
|
|
|
224 retVal = fread(&(chromosomes[i]->lw_cnt), sizeof(chromosomes[i]->lw_cnt), 1, config);
|
|
|
225 retVal = fread(&(chromosomes[i]->sw_cnt), sizeof(chromosomes[i]->sw_cnt), 1, config);
|
|
|
226 retVal = fread(&(chromosomes[i]->cw_cnt), sizeof(chromosomes[i]->cw_cnt), 1, config);
|
|
|
227
|
|
|
228 /* create windows structures */
|
|
|
229
|
|
|
230
|
|
|
231 chromosomes[i]->lw = (struct window *) malloc (sizeof(struct window) * chromosomes[i]->lw_cnt);
|
|
|
232 chromosomes[i]->sw = (struct window *) malloc (sizeof(struct window) * chromosomes[i]->sw_cnt);
|
|
|
233 chromosomes[i]->cw = (struct window *) malloc (sizeof(struct window) * chromosomes[i]->cw_cnt);
|
|
|
234
|
|
|
235
|
|
|
236
|
|
|
237 /* long windows */
|
|
|
238 for (wcnt=0; wcnt<chromosomes[i]->lw_cnt; wcnt++){
|
|
|
239 retVal = fread(&(chromosomes[i]->lw[wcnt].start), sizeof(chromosomes[i]->lw[wcnt].start), 1, config);
|
|
|
240 retVal = fread(&(chromosomes[i]->lw[wcnt].end), sizeof(chromosomes[i]->lw[wcnt].end), 1, config);
|
|
|
241 retVal = fread(&(chromosomes[i]->lw[wcnt].gc), sizeof(chromosomes[i]->lw[wcnt].gc), 1, config);
|
|
|
242 // this is the default, auto-control-picker will fix this
|
|
|
243 chromosomes[i]->lw[wcnt].isControl = 1;
|
|
|
244 chromosomes[i]->lw[wcnt].depth = 0.0;
|
|
|
245 }
|
|
|
246
|
|
|
247 /* short windows */
|
|
|
248 for (wcnt=0; wcnt<chromosomes[i]->sw_cnt; wcnt++){
|
|
|
249 retVal = fread(&(chromosomes[i]->sw[wcnt].start), sizeof(chromosomes[i]->sw[wcnt].start), 1, config);
|
|
|
250 retVal = fread(&(chromosomes[i]->sw[wcnt].end), sizeof(chromosomes[i]->sw[wcnt].end), 1, config);
|
|
|
251 retVal = fread(&(chromosomes[i]->sw[wcnt].gc), sizeof(chromosomes[i]->sw[wcnt].gc), 1, config);
|
|
|
252 chromosomes[i]->sw[wcnt].isControl = 1;
|
|
|
253 chromosomes[i]->sw[wcnt].depth = 0.0;
|
|
|
254 }
|
|
|
255
|
|
|
256 /* copy windows */
|
|
|
257 for (wcnt=0; wcnt<chromosomes[i]->cw_cnt; wcnt++){
|
|
|
258 retVal = fread(&(chromosomes[i]->cw[wcnt].start), sizeof(chromosomes[i]->cw[wcnt].start), 1, config);
|
|
|
259 retVal = fread(&(chromosomes[i]->cw[wcnt].end), sizeof(chromosomes[i]->cw[wcnt].end), 1, config);
|
|
|
260 retVal = fread(&(chromosomes[i]->cw[wcnt].gc), sizeof(chromosomes[i]->cw[wcnt].gc), 1, config);
|
|
|
261 chromosomes[i]->cw[wcnt].isControl = 1;
|
|
|
262 chromosomes[i]->cw[wcnt].depth = 0.0;
|
|
|
263 }
|
|
|
264
|
|
|
265 /* fill in the rest of the chromosomes structure */
|
|
|
266
|
|
|
267 }
|
|
|
268
|
|
|
269 fprintf(stdout, "[OK]. %d chromosomes loaded.\n", num_chrom);
|
|
|
270
|
|
|
271 fclose(config);
|
|
|
272 }
|
|
|
273
|
|
|
274
|
|
|
275
|
|
|
276 static int compChr(const void *p1, const void *p2){
|
|
|
277 /* compare function to sort the chromosome pointer array */
|
|
|
278 struct chrom *a, *b;
|
|
|
279
|
|
|
280 a = *((struct chrom **)p1);
|
|
|
281 b = *((struct chrom **)p2);
|
|
|
282
|
|
|
283
|
|
|
284 return (strcmp ( a->name, b->name) );
|
|
|
285
|
|
|
286 }
|
|
|
287
|
|
|
288
|
|
|
289 int endswith(char *src, char *end){
|
|
|
290
|
|
|
291
|
|
|
292 int endlen = strlen(end);
|
|
|
293 int srclen = strlen(src);
|
|
|
294 int copyindex;
|
|
|
295
|
|
|
296
|
|
|
297 if (endlen > srclen)
|
|
|
298 return 0;
|
|
|
299
|
|
|
300 copyindex = srclen - endlen;
|
|
|
301
|
|
|
302 if (memcmp(src+copyindex, end, endlen))
|
|
|
303 return 0;
|
|
|
304 else
|
|
|
305 return 1;
|
|
|
306
|
|
|
307
|
|
|
308 }
|
|
|
309
|
|
|
310 void trimspace(char *str){
|
|
|
311 int len;
|
|
|
312 len = strlen(str) - 1;
|
|
|
313 while(1){
|
|
|
314 if (isspace(str[len]))
|
|
|
315 str[len--]=0;
|
|
|
316 else
|
|
|
317 break;
|
|
|
318 }
|
|
|
319 }
|
