bam_pileup.c

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