bam_pileup.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <ctype.h>
   4 #include <assert.h>
   5 #include "sam.h"
   6
   7 typedef struct __linkbuf_t {
   8         bam1_t b;
   9         uint32_t beg, end;
  10         struct __linkbuf_t *next;
  11 } lbnode_t;
  12
  13 /* --- BEGIN: Memory pool */
  14
  15 typedef struct {
  16         int cnt, n, max;
  17         lbnode_t **buf;
  18 } mempool_t;
  19
  20 static mempool_t *mp_init()
  21 {
  22         mempool_t *mp;
  23         mp = (mempool_t*)calloc(1, sizeof(mempool_t));
  24         return mp;
  25 }
  26 static void mp_destroy(mempool_t *mp)
  27 {
  28         int k;
  29         for (k = 0; k < mp->n; ++k) {
  30                 free(mp->buf[k]->b.data);
  31                 free(mp->buf[k]);
  32         }
  33         free(mp->buf);
  34         free(mp);
  35 }
  36 static inline lbnode_t *mp_alloc(mempool_t *mp)
  37 {
  38         ++mp->cnt;
  39         if (mp->n == 0) return (lbnode_t*)calloc(1, sizeof(lbnode_t));
  40         else return mp->buf[--mp->n];
  41 }
  42 static inline void mp_free(mempool_t *mp, lbnode_t *p)
  43 {
  44         --mp->cnt; p->next = 0; // clear lbnode_t::next here
  45         if (mp->n == mp->max) {
  46                 mp->max = mp->max? mp->max<<1 : 256;
  47                 mp->buf = (lbnode_t**)realloc(mp->buf, sizeof(lbnode_t*) * mp->max);
  48         }
  49         mp->buf[mp->n++] = p;
  50 }
  51
  52 /* --- END: Memory pool */
  53
  54 /* --- BEGIN: Auxiliary functions */
  55
  56 static inline int resolve_cigar(bam_pileup1_t *p, uint32_t pos)
  57 {
  58         unsigned k;
  59         bam1_t *b = p->b;
  60         bam1_core_t *c = &b->core;
  61         uint32_t x = c->pos, y = 0;
  62         int ret = 1, is_restart = 1;
  63
  64         if (c->flag&BAM_FUNMAP) return 0; // unmapped read
  65         assert(x <= pos); // otherwise a bug
  66         p->qpos = -1; p->indel = 0; p->is_del = p->is_head = p->is_tail = 0;
  67         for (k = 0; k < c->n_cigar; ++k) {
  68                 int op = bam1_cigar(b)[k] & BAM_CIGAR_MASK; // operation
  69                 int l = bam1_cigar(b)[k] >> BAM_CIGAR_SHIFT; // length
  70                 if (op == BAM_CMATCH) { // NOTE: this assumes the first and the last operation MUST BE a match or a clip
  71                         if (x + l > pos) { // overlap with pos
  72                                 p->indel = p->is_del = 0;
  73                                 p->qpos = y + (pos - x);
  74                                 if (x == pos && is_restart) p->is_head = 1;
  75                                 if (x + l - 1 == pos) { // come to the end of a match
  76                                         if (k < c->n_cigar - 1) { // there are additional operation(s)
  77                                                 uint32_t cigar = bam1_cigar(b)[k+1]; // next CIGAR
  78                                                 int op_next = cigar&BAM_CIGAR_MASK; // next CIGAR operation
  79                                                 if (op_next == BAM_CDEL) p->indel = -(int32_t)(cigar>>BAM_CIGAR_SHIFT); // del
  80                                                 else if (op_next == BAM_CINS) p->indel = cigar>>BAM_CIGAR_SHIFT; // ins
  81                                                 if (op_next == BAM_CDEL || op_next == BAM_CINS) {
  82                                                         if (k + 2 < c->n_cigar) op_next = bam1_cigar(b)[k+2]&BAM_CIGAR_MASK;
  83                                                         else p->is_tail = 1;
  84                                                 }
  85                                                 if (op_next == BAM_CSOFT_CLIP || op_next == BAM_CREF_SKIP || op_next == BAM_CHARD_CLIP)
  86                                                         p->is_tail = 1; // tail
  87                                         } else p->is_tail = 1; // this is the last operation; set tail
  88                                 }
  89                         }
  90                         x += l; y += l;
  91                 } else if (op == BAM_CDEL) { // then set ->is_del
  92                         if (x + l > pos) {
  93                                 p->indel = 0; p->is_del = 1;
  94                                 p->qpos = y + (pos - x);
  95                         }
  96                         x += l;
  97                 } else if (op == BAM_CREF_SKIP) x += l;
  98                 else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
  99                 is_restart = (op == BAM_CREF_SKIP || op == BAM_CSOFT_CLIP || op == BAM_CHARD_CLIP);
 100                 if (x > pos) {
 101                         if (op == BAM_CREF_SKIP) ret = 0; // then do not put it into pileup at all
 102                         break;
 103                 }
 104         }
 105         assert(x > pos); // otherwise a bug
 106         return ret;
 107 }
 108
 109 /* --- END: Auxiliary functions */
 110
 111 struct __bam_plbuf_t {
 112         mempool_t *mp;
 113         lbnode_t *head, *tail, *dummy;
 114         bam_pileup_f func;
 115         void *func_data;
 116         int32_t tid, pos, max_tid, max_pos;
 117         int max_pu, is_eof;
 118         bam_pileup1_t *pu;
 119         int flag_mask;
 120 };
 121
 122 void bam_plbuf_reset(bam_plbuf_t *buf)
 123 {
 124         lbnode_t *p, *q;
 125         buf->max_tid = buf->max_pos = -1;
 126         buf->tid = buf->pos = 0;
 127         buf->is_eof = 0;
 128         for (p = buf->head; p->next;) {
 129                 q = p->next;
 130                 mp_free(buf->mp, p);
 131                 p = q;
 132         }
 133         buf->head = buf->tail;
 134 }
 135
 136 void bam_plbuf_set_mask(bam_plbuf_t *buf, int mask)
 137 {
 138         if (mask < 0) buf->flag_mask = BAM_DEF_MASK;
 139         else buf->flag_mask = BAM_FUNMAP | mask;
 140 }
 141
 142 bam_plbuf_t *bam_plbuf_init(bam_pileup_f func, void *data)
 143 {
 144         bam_plbuf_t *buf;
 145         buf = (bam_plbuf_t*)calloc(1, sizeof(bam_plbuf_t));
 146         buf->func = func; buf->func_data = data;
 147         buf->mp = mp_init();
 148         buf->head = buf->tail = mp_alloc(buf->mp);
 149         buf->dummy = mp_alloc(buf->mp);
 150         buf->max_tid = buf->max_pos = -1;
 151         buf->flag_mask = BAM_DEF_MASK;
 152         return buf;
 153 }
 154
 155 void bam_plbuf_destroy(bam_plbuf_t *buf)
 156 {
 157         mp_free(buf->mp, buf->dummy);
 158         mp_free(buf->mp, buf->head);
 159         if (buf->mp->cnt != 0)
 160                 fprintf(stderr, "[bam_plbuf_destroy] memory leak: %d. Continue anyway.\n", buf->mp->cnt);
 161         mp_destroy(buf->mp);
 162         free(buf->pu);
 163         free(buf);
 164 }
 165
 166 int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf)
 167 {
 168         if (b) { // fill buffer
 169                 if (b->core.tid < 0) return 0;
 170                 if (b->core.flag & buf->flag_mask) return 0;
 171                 bam_copy1(&buf->tail->b, b);
 172                 buf->tail->beg = b->core.pos; buf->tail->end = bam_calend(&b->core, bam1_cigar(b));
 173                 if (b->core.tid < buf->max_tid) {
 174                         fprintf(stderr, "[bam_pileup_core] the input is not sorted (chromosomes out of order)\n");
 175                         return -1;
 176                 }
 177                 if ((b->core.tid == buf->max_tid) && (buf->tail->beg < buf->max_pos)) {
 178                         fprintf(stderr, "[bam_pileup_core] the input is not sorted (reads out of order)\n");
 179                         return -1;
 180                 }
 181                 buf->max_tid = b->core.tid; buf->max_pos = buf->tail->beg;
 182                 if (buf->tail->end > buf->pos || buf->tail->b.core.tid > buf->tid) {
 183                         buf->tail->next = mp_alloc(buf->mp);
 184                         buf->tail = buf->tail->next;
 185                 }
 186         } else buf->is_eof = 1;
 187         while (buf->is_eof || buf->max_tid > buf->tid || (buf->max_tid == buf->tid && buf->max_pos > buf->pos)) {
 188                 int n_pu = 0;
 189                 lbnode_t *p, *q;
 190                 buf->dummy->next = buf->head;
 191                 for (p = buf->head, q = buf->dummy; p->next; q = p, p = p->next) {
 192                         if (p->b.core.tid < buf->tid || (p->b.core.tid == buf->tid && p->end <= buf->pos)) { // then remove from the list
 193                                 q->next = p->next; mp_free(buf->mp, p); p = q;
 194                         } else if (p->b.core.tid == buf->tid && p->beg <= buf->pos) { // here: p->end > pos; then add to pileup
 195                                 if (n_pu == buf->max_pu) { // then double the capacity
 196                                         buf->max_pu = buf->max_pu? buf->max_pu<<1 : 256;
 197                                         buf->pu = (bam_pileup1_t*)realloc(buf->pu, sizeof(bam_pileup1_t) * buf->max_pu);
 198                                 }
 199                                 buf->pu[n_pu].b = &p->b;
 200                                 if (resolve_cigar(buf->pu + n_pu, buf->pos)) ++n_pu; // skip the read if we are looking at BAM_CREF_SKIP
 201                         }
 202                 }
 203                 buf->head = buf->dummy->next; // dummy->next may be changed
 204                 if (n_pu) { // then call user defined function
 205                         buf->func(buf->tid, buf->pos, n_pu, buf->pu, buf->func_data);
 206                 }
 207                 // update tid and pos
 208                 if (buf->head->next) {
 209                         if (buf->tid > buf->head->b.core.tid) {
 210                                 fprintf(stderr, "[bam_plbuf_push] unsorted input. Pileup aborts.\n");
 211                                 return 1;
 212                         }
 213                 }
 214                 if (buf->tid < buf->head->b.core.tid) { // come to a new reference sequence
 215                         buf->tid = buf->head->b.core.tid; buf->pos = buf->head->beg; // jump to the next reference
 216                 } else if (buf->pos < buf->head->beg) { // here: tid == head->b.core.tid
 217                         buf->pos = buf->head->beg; // jump to the next position
 218                 } else ++buf->pos; // scan contiguously
 219                 if (buf->is_eof && buf->head->next == 0) break;
 220         }
 221         return 0;
 222 }
 223
 224 int bam_pileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data)
 225 {
 226         bam_plbuf_t *buf;
 227         int ret;
 228         bam1_t *b;
 229         b = bam_init1();
 230         buf = bam_plbuf_init(func, func_data);
 231         bam_plbuf_set_mask(buf, mask);
 232         while ((ret = bam_read1(fp, b)) >= 0)
 233                 bam_plbuf_push(b, buf);
 234         bam_plbuf_push(0, buf);
 235         bam_plbuf_destroy(buf);
 236         bam_destroy1(b);
 237         return 0;
 238 }