bwamem.c

/* The MIT License

   Copyright (c) 2018-     Dana-Farber Cancer Institute
                 2009-2018 Broad Institute, Inc.
                 2008-2009 Genome Research Ltd. (GRL)

   Permission is hereby granted, free of charge, to any person obtaining
   a copy of this software and associated documentation files (the
   "Software"), to deal in the Software without restriction, including
   without limitation the rights to use, copy, modify, merge, publish,
   distribute, sublicense, and/or sell copies of the Software, and to
   permit persons to whom the Software is furnished to do so, subject to
   the following conditions:

   The above copyright notice and this permission notice shall be
   included in all copies or substantial portions of the Software.

   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
   EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
   MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
   NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
   BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
   ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
   CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
   SOFTWARE.
*/
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#include <assert.h>
#include <limits.h>
#include <math.h>
#ifdef HAVE_PTHREAD
#include <pthread.h>
#endif

#include "kstring.h"
#include "bwamem.h"
#include "bntseq.h"
#include "ksw.h"
#include "kvec.h"
#include "ksort.h"
#include "utils.h"

#ifdef USE_MALLOC_WRAPPERS
#  include "malloc_wrap.h"
#endif

/* Theory on probability and scoring *ungapped* alignment
 *
 * s'(a,b) = log[P(b|a)/P(b)] = log[4P(b|a)], assuming uniform base distribution
 * s'(a,a) = log(4), s'(a,b) = log(4e/3), where e is the error rate
 *
 * Scale s'(a,b) to s(a,a) s.t. s(a,a)=x. Then s(a,b) = x*s'(a,b)/log(4), or conversely: s'(a,b)=s(a,b)*log(4)/x
 *
 * If the matching score is x and mismatch penalty is -y, we can compute error rate e:
 *   e = .75 * exp[-log(4) * y/x]
 *
 * log P(seq) = \sum_i log P(b_i|a_i) = \sum_i {s'(a,b) - log(4)}
 *   = \sum_i { s(a,b)*log(4)/x - log(4) } = log(4) * (S/x - l)
 *
 * where S=\sum_i s(a,b) is the alignment score. Converting to the phred scale:
 *   Q(seq) = -10/log(10) * log P(seq) = 10*log(4)/log(10) * (l - S/x) = 6.02 * (l - S/x)
 *
 *
 * Gap open (zero gap): q' = log[P(gap-open)], r' = log[P(gap-ext)] (see Durbin et al. (1998) Section 4.1)
 * Then q = x*log[P(gap-open)]/log(4), r = x*log[P(gap-ext)]/log(4)
 *
 * When there are gaps, l should be the length of alignment matches (i.e. the M operator in CIGAR)
 */

static const bntseq_t *global_bns = 0; // for debugging only

mem_opt_t *mem_opt_init()
{
	mem_opt_t *o;
	o = calloc(1, sizeof(mem_opt_t));
	o->flag = 0;
	o->a = 1; o->b = 4;
	o->o_del = o->o_ins = 6;
	o->e_del = o->e_ins = 1;
	o->w = 100;
	o->T = 30;
	o->zdrop = 100;
	o->pen_unpaired = 17;
	o->pen_clip5 = o->pen_clip3 = 5;

	o->max_mem_intv = 20;

	o->min_seed_len = 19;
	o->split_width = 10;
	o->max_occ = 500;
	o->max_chain_gap = 10000;
	o->max_ins = 10000;
	o->mask_level = 0.50;
	o->drop_ratio = 0.50;
	o->XA_drop_ratio = 0.80;
	o->split_factor = 1.5;
	o->chunk_size = 30000000;
	o->n_threads = 1;
	o->max_XA_hits = 5;
	o->max_XA_hits_alt = 200;
	o->max_matesw = 50;
	o->mask_level_redun = 0.95;
	o->min_chain_weight = 0;
	o->max_chain_extend = 1<<30;
	o->mapQ_coef_len = 50; o->mapQ_coef_fac = log(o->mapQ_coef_len);
	bwa_fill_scmat(o->a, o->b, o->mat);
	return o;
}

/***************************
 * Collection SA invervals *
 ***************************/

#define intv_lt(a, b) ((a).info < (b).info)
KSORT_INIT(mem_intv, bwtintv_t, intv_lt)


static smem_aux_t *smem_aux_init()
{
	smem_aux_t *a;
	a = calloc(1, sizeof(smem_aux_t));
	a->tmpv[0] = calloc(1, sizeof(bwtintv_v));
	a->tmpv[1] = calloc(1, sizeof(bwtintv_v));
	return a;
}

static void smem_aux_destroy(smem_aux_t *a)
{	
	free(a->tmpv[0]->a); free(a->tmpv[0]);
	free(a->tmpv[1]->a); free(a->tmpv[1]);
	free(a->mem.a); free(a->mem1.a);
	free(a);
}

static void mem_collect_intv(const mem_opt_t *opt, const bwt_t *bwt, int len, const uint8_t *seq, smem_aux_t *a)
{
	int i, k, x = 0, old_n;
	int start_width = 1;
	int split_len = (int)(opt->min_seed_len * opt->split_factor + .499);
	a->mem.n = 0;
	// first pass: find all SMEMs
	while (x < len) {
		if (seq[x] < 4) {
			x = bwt_smem1(bwt, len, seq, x, start_width, &a->mem1, a->tmpv);
			for (i = 0; i < a->mem1.n; ++i) {
				bwtintv_t *p = &a->mem1.a[i];
				int slen = (uint32_t)p->info - (p->info>>32); // seed length
				if (slen >= opt->min_seed_len)
					kv_push(bwtintv_t, a->mem, *p);
			}
		} else ++x;
	}
	// second pass: find MEMs inside a long SMEM
	old_n = a->mem.n;
	for (k = 0; k < old_n; ++k) {
		bwtintv_t *p = &a->mem.a[k];
		int start = p->info>>32, end = (int32_t)p->info;
		if (end - start < split_len || p->x[2] > opt->split_width) continue;
		bwt_smem1(bwt, len, seq, (start + end)>>1, p->x[2]+1, &a->mem1, a->tmpv);
		for (i = 0; i < a->mem1.n; ++i)
			if ((uint32_t)a->mem1.a[i].info - (a->mem1.a[i].info>>32) >= opt->min_seed_len)
				kv_push(bwtintv_t, a->mem, a->mem1.a[i]);
	}
	// third pass: LAST-like
	if (opt->max_mem_intv > 0) {
		x = 0;
		while (x < len) {
			if (seq[x] < 4) {
				if (1) {
					bwtintv_t m;
					x = bwt_seed_strategy1(bwt, len, seq, x, opt->min_seed_len, opt->max_mem_intv, &m);
					if (m.x[2] > 0) kv_push(bwtintv_t, a->mem, m);
				} else { // for now, we never come to this block which is slower
					x = bwt_smem1a(bwt, len, seq, x, start_width, opt->max_mem_intv, &a->mem1, a->tmpv);
					for (i = 0; i < a->mem1.n; ++i)
						kv_push(bwtintv_t, a->mem, a->mem1.a[i]);
				}
			} else ++x;
		}
	}
	// sort
	ks_introsort(mem_intv, a->mem.n, a->mem.a);
}

/************
 * Chaining *
 ************/

#include "kbtree.h"

#define chain_cmp(a, b) (((b).pos < (a).pos) - ((a).pos < (b).pos))
KBTREE_INIT(chn, mem_chain_t, chain_cmp)

// return 1 if the seed is merged into the chain
static int test_and_merge(const mem_opt_t *opt, int64_t l_pac, mem_chain_t *c, const mem_seed_t *p, int seed_rid)
{
	int64_t qend, rend, x, y;
	const mem_seed_t *last = &c->seeds[c->n-1];
	qend = last->qbeg + last->len;
	rend = last->rbeg + last->len;
	if (seed_rid != c->rid) return 0; // different chr; request a new chain
	if (p->qbeg >= c->seeds[0].qbeg && p->qbeg + p->len <= qend && p->rbeg >= c->seeds[0].rbeg && p->rbeg + p->len <= rend)
		return 1; // contained seed; do nothing
	if ((last->rbeg < l_pac || c->seeds[0].rbeg < l_pac) && p->rbeg >= l_pac) return 0; // don't chain if on different strand
	x = p->qbeg - last->qbeg; // always non-negtive
	y = p->rbeg - last->rbeg;
	if (y >= 0 && x - y <= opt->w && y - x <= opt->w && x - last->len < opt->max_chain_gap && y - last->len < opt->max_chain_gap) { // grow the chain
		if (c->n == c->m) {
			c->m <<= 1;
			c->seeds = realloc(c->seeds, c->m * sizeof(mem_seed_t));
		}
		c->seeds[c->n++] = *p;
		return 1;
	}
	return 0; // request to add a new chain
}

int mem_chain_weight(const mem_chain_t *c)
{
	int64_t end;
	int j, w = 0, tmp;
	for (j = 0, end = 0; j < c->n; ++j) {
		const mem_seed_t *s = &c->seeds[j];
		if (s->qbeg >= end) w += s->len;
		else if (s->qbeg + s->len > end) w += s->qbeg + s->len - end;
		end = end > s->qbeg + s->len? end : s->qbeg + s->len;
	}
	tmp = w; w = 0;
	for (j = 0, end = 0; j < c->n; ++j) {
		const mem_seed_t *s = &c->seeds[j];
		if (s->rbeg >= end) w += s->len;
		else if (s->rbeg + s->len > end) w += s->rbeg + s->len - end;
		end = end > s->rbeg + s->len? end : s->rbeg + s->len;
	}
	w = w < tmp? w : tmp;
	return w < 1<<30? w : (1<<30)-1;
}

void mem_print_chain(const bntseq_t *bns, mem_chain_v *chn)
{
	int i, j;
	for (i = 0; i < chn->n; ++i) {
		mem_chain_t *p = &chn->a[i];
		err_printf("* Found CHAIN(%d): n=%d; weight=%d", i, p->n, mem_chain_weight(p));
		for (j = 0; j < p->n; ++j) {
			bwtint_t pos;
			int is_rev;
			pos = bns_depos(bns, p->seeds[j].rbeg, &is_rev);
			if (is_rev) pos -= p->seeds[j].len - 1;
			err_printf("\t%d;%d;%d,%ld(%s:%c%ld)", p->seeds[j].score, p->seeds[j].len, p->seeds[j].qbeg, (long)p->seeds[j].rbeg, bns->anns[p->rid].name, "+-"[is_rev], (long)(pos - bns->anns[p->rid].offset) + 1);
		}
		err_putchar('\n');
	}
}

mem_chain_v mem_chain(const mem_opt_t *opt, const bwt_t *bwt, const bntseq_t *bns, int len, const uint8_t *seq, void *buf)
{
	int i, b, e, l_rep;
	int64_t l_pac = bns->l_pac;
	mem_chain_v chain;
	kbtree_t(chn) *tree;
	smem_aux_t *aux;

	kv_init(chain);
	if (len < opt->min_seed_len) return chain; // if the query is shorter than the seed length, no match
	tree = kb_init(chn, KB_DEFAULT_SIZE);

	aux = buf? (smem_aux_t*)buf : smem_aux_init();
	mem_collect_intv(opt, bwt, len, seq, aux);
	for (i = 0, b = e = l_rep = 0; i < aux->mem.n; ++i) { // compute frac_rep
		bwtintv_t *p = &aux->mem.a[i];
		int sb = (p->info>>32), se = (uint32_t)p->info;
		if (p->x[2] <= opt->max_occ) continue;
		if (sb > e) l_rep += e - b, b = sb, e = se;
		else e = e > se? e : se;
	}
	l_rep += e - b;
	for (i = 0; i < aux->mem.n; ++i) {
		bwtintv_t *p = &aux->mem.a[i];
		int step, count, slen = (uint32_t)p->info - (p->info>>32); // seed length
		int64_t k;
		// if (slen < opt->min_seed_len) continue; // ignore if too short or too repetitive
		step = p->x[2] > opt->max_occ? p->x[2] / opt->max_occ : 1;
		for (k = count = 0; k < p->x[2] && count < opt->max_occ; k += step, ++count) {
			mem_chain_t tmp, *lower, *upper;
			mem_seed_t s;
			int rid, to_add = 0;
			s.rbeg = tmp.pos = bwt_sa(bwt, p->x[0] + k); // this is the base coordinate in the forward-reverse reference
			s.qbeg = p->info>>32;
			s.score= s.len = slen;
			rid = bns_intv2rid(bns, s.rbeg, s.rbeg + s.len);
			if (rid < 0) continue; // bridging multiple reference sequences or the forward-reverse boundary; TODO: split the seed; don't discard it!!!
			if (kb_size(tree)) {
				kb_intervalp(chn, tree, &tmp, &lower, &upper); // find the closest chain
				if (!lower || !test_and_merge(opt, l_pac, lower, &s, rid)) to_add = 1;
			} else to_add = 1;
			if (to_add) { // add the seed as a new chain
				tmp.n = 1; tmp.m = 4;
				tmp.seeds = calloc(tmp.m, sizeof(mem_seed_t));
				tmp.seeds[0] = s;
				tmp.rid = rid;
				tmp.is_alt = !!bns->anns[rid].is_alt;
				kb_putp(chn, tree, &tmp);
			}
		}
	}
	if (buf == 0) smem_aux_destroy(aux);

	kv_resize(mem_chain_t, chain, kb_size(tree));

	#define traverse_func(p_) (chain.a[chain.n++] = *(p_))
	__kb_traverse(mem_chain_t, tree, traverse_func);
	#undef traverse_func

	for (i = 0; i < chain.n; ++i) chain.a[i].frac_rep = (float)l_rep / len;
	if (bwa_verbose >= 4) printf("* fraction of repetitive seeds: %.3f\n", (float)l_rep / len);

	kb_destroy(chn, tree);
	return chain;
}

/********************
 * Filtering chains *
 ********************/

#define chn_beg(ch) ((ch).seeds->qbeg)
#define chn_end(ch) ((ch).seeds[(ch).n-1].qbeg + (ch).seeds[(ch).n-1].len)

#define flt_lt(a, b) ((a).w > (b).w)
KSORT_INIT(mem_flt, mem_chain_t, flt_lt)

int mem_chain_flt(const mem_opt_t *opt, int n_chn, mem_chain_t *a)
{
	int i, k;
	kvec_t(int) chains = {0,0,0}; // this keeps int indices of the non-overlapping chains
	if (n_chn == 0) return 0; // no need to filter
	// compute the weight of each chain and drop chains with small weight
	for (i = k = 0; i < n_chn; ++i) {
		mem_chain_t *c = &a[i];
		c->first = -1; c->kept = 0;
		c->w = mem_chain_weight(c);
		if (c->w < opt->min_chain_weight) free(c->seeds);
		else a[k++] = *c;
	}
	n_chn = k;
	ks_introsort(mem_flt, n_chn, a);
	// pairwise chain comparisons
	a[0].kept = 3;
	kv_push(int, chains, 0);
	for (i = 1; i < n_chn; ++i) {
		int large_ovlp = 0;
		for (k = 0; k < chains.n; ++k) {
			int j = chains.a[k];
			int b_max = chn_beg(a[j]) > chn_beg(a[i])? chn_beg(a[j]) : chn_beg(a[i]);
			int e_min = chn_end(a[j]) < chn_end(a[i])? chn_end(a[j]) : chn_end(a[i]);
			if (e_min > b_max && (!a[j].is_alt || a[i].is_alt)) { // have overlap; don't consider ovlp where the kept chain is ALT while the current chain is primary
				int li = chn_end(a[i]) - chn_beg(a[i]);
				int lj = chn_end(a[j]) - chn_beg(a[j]);
				int min_l = li < lj? li : lj;
				if (e_min - b_max >= min_l * opt->mask_level && min_l < opt->max_chain_gap) { // significant overlap
					large_ovlp = 1;
					if (a[j].first < 0) a[j].first = i; // keep the first shadowed hit s.t. mapq can be more accurate
					if (a[i].w < a[j].w * opt->drop_ratio && a[j].w - a[i].w >= opt->min_seed_len<<1)
						break;
				}
			}
		}
		if (k == chains.n) {
			kv_push(int, chains, i);
			a[i].kept = large_ovlp? 2 : 3;
		}
	}
	for (i = 0; i < chains.n; ++i) {
		mem_chain_t *c = &a[chains.a[i]];
		if (c->first >= 0) a[c->first].kept = 1;
	}
	free(chains.a);
	for (i = k = 0; i < n_chn; ++i) { // don't extend more than opt->max_chain_extend .kept=1/2 chains
		if (a[i].kept == 0 || a[i].kept == 3) continue;
		if (++k >= opt->max_chain_extend) break;
	}
	for (; i < n_chn; ++i)
		if (a[i].kept < 3) a[i].kept = 0;
	for (i = k = 0; i < n_chn; ++i) { // free discarded chains
		mem_chain_t *c = &a[i];
		if (c->kept == 0) free(c->seeds);
		else a[k++] = a[i];
	}
	return k;
}

/******************************
 * De-overlap single-end hits *
 ******************************/

#define alnreg_slt2(a, b) ((a).re < (b).re)
KSORT_INIT(mem_ars2, mem_alnreg_t, alnreg_slt2)

#define alnreg_slt(a, b) ((a).score > (b).score || ((a).score == (b).score && ((a).rb < (b).rb || ((a).rb == (b).rb && (a).qb < (b).qb))))
KSORT_INIT(mem_ars, mem_alnreg_t, alnreg_slt)

#define alnreg_hlt(a, b)  ((a).score > (b).score || ((a).score == (b).score && ((a).is_alt < (b).is_alt || ((a).is_alt == (b).is_alt && (a).hash < (b).hash))))
KSORT_INIT(mem_ars_hash, mem_alnreg_t, alnreg_hlt)

#define alnreg_hlt2(a, b) ((a).is_alt < (b).is_alt || ((a).is_alt == (b).is_alt && ((a).score > (b).score || ((a).score == (b).score && (a).hash < (b).hash))))
KSORT_INIT(mem_ars_hash2, mem_alnreg_t, alnreg_hlt2)

#define PATCH_MAX_R_BW 0.05f
#define PATCH_MIN_SC_RATIO 0.90f

int mem_patch_reg(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, uint8_t *query, const mem_alnreg_t *a, const mem_alnreg_t *b, int *_w)
{
	int w, score, q_s, r_s;
	double r;
	if (bns == 0 || pac == 0 || query == 0) return 0;
	assert(a->rid == b->rid && a->rb <= b->rb);
	if (a->rb < bns->l_pac && b->rb >= bns->l_pac) return 0; // on different strands
	if (a->qb >= b->qb || a->qe >= b->qe || a->re >= b->re) return 0; // not colinear
	w = (a->re - b->rb) - (a->qe - b->qb); // required bandwidth
	w = w > 0? w : -w; // l = abs(l)
	r = (double)(a->re - b->rb) / (b->re - a->rb) - (double)(a->qe - b->qb) / (b->qe - a->qb); // relative bandwidth
	r = r > 0.? r : -r; // r = fabs(r)
	if (bwa_verbose >= 4)
		printf("* potential hit merge between [%d,%d)<=>[%ld,%ld) and [%d,%d)<=>[%ld,%ld), @ %s; w=%d, r=%.4g\n",
			   a->qb, a->qe, (long)a->rb, (long)a->re, b->qb, b->qe, (long)b->rb, (long)b->re, bns->anns[a->rid].name, w, r);
	if (a->re < b->rb || a->qe < b->qb) { // no overlap on query or on ref
		if (w > opt->w<<1 || r >= PATCH_MAX_R_BW) return 0; // the bandwidth or the relative bandwidth is too large
	} else if (w > opt->w<<2 || r >= PATCH_MAX_R_BW*2) return 0; // more permissive if overlapping on both ref and query
	// global alignment
	w += a->w + b->w;
	w = w < opt->w<<2? w : opt->w<<2;
	if (bwa_verbose >= 4) printf("* test potential hit merge with global alignment; w=%d\n", w);
	bwa_gen_cigar2(opt->mat, opt->o_del, opt->e_del, opt->o_ins, opt->e_ins, w, bns->l_pac, pac, b->qe - a->qb, query + a->qb, a->rb, b->re, &score, 0, 0);
	q_s = (int)((double)(b->qe - a->qb) / ((b->qe - b->qb) + (a->qe - a->qb)) * (b->score + a->score) + .499); // predicted score from query
	r_s = (int)((double)(b->re - a->rb) / ((b->re - b->rb) + (a->re - a->rb)) * (b->score + a->score) + .499); // predicted score from ref
	if (bwa_verbose >= 4) printf("* score=%d;(%d,%d)\n", score, q_s, r_s);
	if ((double)score / (q_s > r_s? q_s : r_s) < PATCH_MIN_SC_RATIO) return 0;
	*_w = w;
	return score;
}

int mem_sort_dedup_patch(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, uint8_t *query, int n, mem_alnreg_t *a)
{
	int m, i, j;
	if (n <= 1) return n;
	ks_introsort(mem_ars2, n, a); // sort by the END position, not START!
	for (i = 0; i < n; ++i) a[i].n_comp = 1;
	for (i = 1; i < n; ++i) {
		mem_alnreg_t *p = &a[i];
		if (p->rid != a[i-1].rid || p->rb >= a[i-1].re + opt->max_chain_gap) continue; // then no need to go into the loop below
		for (j = i - 1; j >= 0 && p->rid == a[j].rid && p->rb < a[j].re + opt->max_chain_gap; --j) {
			mem_alnreg_t *q = &a[j];
			int64_t or, oq, mr, mq;
			int score, w;
			if (q->qe == q->qb) continue; // a[j] has been excluded
			or = q->re - p->rb; // overlap length on the reference
			oq = q->qb < p->qb? q->qe - p->qb : p->qe - q->qb; // overlap length on the query
			mr = q->re - q->rb < p->re - p->rb? q->re - q->rb : p->re - p->rb; // min ref len in alignment
			mq = q->qe - q->qb < p->qe - p->qb? q->qe - q->qb : p->qe - p->qb; // min qry len in alignment
			if (or > opt->mask_level_redun * mr && oq > opt->mask_level_redun * mq) { // one of the hits is redundant
				if (p->score < q->score) {
					p->qe = p->qb;
					break;
				} else q->qe = q->qb;
			} else if (q->rb < p->rb && (score = mem_patch_reg(opt, bns, pac, query, q, p, &w)) > 0) { // then merge q into p
				p->n_comp += q->n_comp + 1;
				p->seedcov = p->seedcov > q->seedcov? p->seedcov : q->seedcov;
				p->sub = p->sub > q->sub? p->sub : q->sub;
				p->csub = p->csub > q->csub? p->csub : q->csub;
				p->qb = q->qb, p->rb = q->rb;
				p->truesc = p->score = score;
				p->w = w;
				q->qb = q->qe;
			}
		}
	}
	for (i = 0, m = 0; i < n; ++i) // exclude identical hits
		if (a[i].qe > a[i].qb) {
			if (m != i) a[m++] = a[i];
			else ++m;
		}
	n = m;
	ks_introsort(mem_ars, n, a);
	for (i = 1; i < n; ++i) { // mark identical hits
		if (a[i].score == a[i-1].score && a[i].rb == a[i-1].rb && a[i].qb == a[i-1].qb)
			a[i].qe = a[i].qb;
	}
	for (i = 1, m = 1; i < n; ++i) // exclude identical hits
		if (a[i].qe > a[i].qb) {
			if (m != i) a[m++] = a[i];
			else ++m;
		}
	return m;
}

typedef kvec_t(int) int_v;

static void mem_mark_primary_se_core(const mem_opt_t *opt, int n, mem_alnreg_t *a, int_v *z)
{ // similar to the loop in mem_chain_flt()
	int i, k, tmp;
	tmp = opt->a + opt->b;
	tmp = opt->o_del + opt->e_del > tmp? opt->o_del + opt->e_del : tmp;
	tmp = opt->o_ins + opt->e_ins > tmp? opt->o_ins + opt->e_ins : tmp;
	z->n = 0;
	kv_push(int, *z, 0);
	for (i = 1; i < n; ++i) {
		for (k = 0; k < z->n; ++k) {
			int j = z->a[k];
			int b_max = a[j].qb > a[i].qb? a[j].qb : a[i].qb;
			int e_min = a[j].qe < a[i].qe? a[j].qe : a[i].qe;
			if (e_min > b_max) { // have overlap
				int min_l = a[i].qe - a[i].qb < a[j].qe - a[j].qb? a[i].qe - a[i].qb : a[j].qe - a[j].qb;
				if (e_min - b_max >= min_l * opt->mask_level) { // significant overlap
					if (a[j].sub == 0) a[j].sub = a[i].score;
					if (a[j].score - a[i].score <= tmp && (a[j].is_alt || !a[i].is_alt))
						++a[j].sub_n;
					break;
				}
			}
		}
		if (k == z->n) kv_push(int, *z, i);
		else a[i].secondary = z->a[k];
	}
}

int mem_mark_primary_se(const mem_opt_t *opt, int n, mem_alnreg_t *a, int64_t id)
{
	int i, n_pri;
	int_v z = {0,0,0};
	if (n == 0) return 0;
	for (i = n_pri = 0; i < n; ++i) {
		a[i].sub = a[i].alt_sc = 0, a[i].secondary = a[i].secondary_all = -1, a[i].hash = hash_64(id+i);
		if (!a[i].is_alt) ++n_pri;
	}
	ks_introsort(mem_ars_hash, n, a);
	mem_mark_primary_se_core(opt, n, a, &z);
	for (i = 0; i < n; ++i) {
		mem_alnreg_t *p = &a[i];
		p->secondary_all = i; // keep the rank in the first round
		if (!p->is_alt && p->secondary >= 0 && a[p->secondary].is_alt)
			p->alt_sc = a[p->secondary].score;
	}
	if (n_pri >= 0 && n_pri < n) {
		kv_resize(int, z, n);
		if (n_pri > 0) ks_introsort(mem_ars_hash2, n, a);
		for (i = 0; i < n; ++i) z.a[a[i].secondary_all] = i;
		for (i = 0; i < n; ++i) {
			if (a[i].secondary >= 0) {
				a[i].secondary_all = z.a[a[i].secondary];
				if (a[i].is_alt) a[i].secondary = INT_MAX;
			} else a[i].secondary_all = -1;
		}
		if (n_pri > 0) { // mark primary for hits to the primary assembly only
			for (i = 0; i < n_pri; ++i) a[i].sub = 0, a[i].secondary = -1;
			mem_mark_primary_se_core(opt, n_pri, a, &z);
		}
	} else {
		for (i = 0; i < n; ++i)
			a[i].secondary_all = a[i].secondary;
	}
	free(z.a);
	return n_pri;
}

/*********************************
 * Test if a seed is good enough *
 *********************************/

#define MEM_SHORT_EXT 50
#define MEM_SHORT_LEN 200

#define MEM_HSP_COEF 1.1f
#define MEM_MINSC_COEF 5.5f
#define MEM_SEEDSW_COEF 0.05f

int mem_seed_sw(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, int l_query, const uint8_t *query, const mem_seed_t *s)
{
	int qb, qe, rid;
	int64_t rb, re, mid, l_pac = bns->l_pac;
	uint8_t *rseq = 0;
	kswr_t x;

	if (s->len >= MEM_SHORT_LEN) return -1; // the seed is longer than the max-extend; no need to do SW
	qb = s->qbeg, qe = s->qbeg + s->len;
	rb = s->rbeg, re = s->rbeg + s->len;
	mid = (rb + re) >> 1;
	qb -= MEM_SHORT_EXT; qb = qb > 0? qb : 0;
	qe += MEM_SHORT_EXT; qe = qe < l_query? qe : l_query;
	rb -= MEM_SHORT_EXT; rb = rb > 0? rb : 0;
	re += MEM_SHORT_EXT; re = re < l_pac<<1? re : l_pac<<1;
	if (rb < l_pac && l_pac < re) {
		if (mid < l_pac) re = l_pac;
		else rb = l_pac;
	}
	if (qe - qb >= MEM_SHORT_LEN || re - rb >= MEM_SHORT_LEN) return -1; // the seed seems good enough; no need to do SW

	rseq = bns_fetch_seq(bns, pac, &rb, mid, &re, &rid);
	x = ksw_align2(qe - qb, (uint8_t*)query + qb, re - rb, rseq, 5, opt->mat, opt->o_del, opt->e_del, opt->o_ins, opt->e_ins, KSW_XSTART, 0);
	free(rseq);
	return x.score;
}

void mem_flt_chained_seeds(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, int l_query, const uint8_t *query, int n_chn, mem_chain_t *a)
{
	double min_l = opt->min_chain_weight? MEM_HSP_COEF * opt->min_chain_weight : MEM_MINSC_COEF * log(l_query);
	int i, j, k, min_HSP_score = (int)(opt->a * min_l + .499);
	if (min_l > MEM_SEEDSW_COEF * l_query) return; // don't run the following for short reads
	for (i = 0; i < n_chn; ++i) {
		mem_chain_t *c = &a[i];
		for (j = k = 0; j < c->n; ++j) {
			mem_seed_t *s = &c->seeds[j];
			s->score = mem_seed_sw(opt, bns, pac, l_query, query, s);
			if (s->score < 0 || s->score >= min_HSP_score) {
				s->score = s->score < 0? s->len * opt->a : s->score;
				c->seeds[k++] = *s;
			}
		}
		c->n = k;
	}
}

/****************************************
 * Construct the alignment from a chain *
 ****************************************/

static inline int cal_max_gap(const mem_opt_t *opt, int qlen)
{
	int l_del = (int)((double)(qlen * opt->a - opt->o_del) / opt->e_del + 1.);
	int l_ins = (int)((double)(qlen * opt->a - opt->o_ins) / opt->e_ins + 1.);
	int l = l_del > l_ins? l_del : l_ins;
	l = l > 1? l : 1;
	return l < opt->w<<1? l : opt->w<<1;
}

#define MAX_BAND_TRY  2

void mem_chain2aln(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, int l_query, const uint8_t *query, const mem_chain_t *c, mem_alnreg_v *av)
{
	int i, k, rid, max_off[2], aw[2]; // aw: actual bandwidth used in extension
	int64_t l_pac = bns->l_pac, rmax[2], tmp, max = 0;
	const mem_seed_t *s;
	uint8_t *rseq = 0;
	uint64_t *srt;

	if (c->n == 0) return;
	// get the max possible span
	rmax[0] = l_pac<<1; rmax[1] = 0;
	for (i = 0; i < c->n; ++i) {
		int64_t b, e;
		const mem_seed_t *t = &c->seeds[i];
		b = t->rbeg - (t->qbeg + cal_max_gap(opt, t->qbeg));
		e = t->rbeg + t->len + ((l_query - t->qbeg - t->len) + cal_max_gap(opt, l_query - t->qbeg - t->len));
		rmax[0] = rmax[0] < b? rmax[0] : b;
		rmax[1] = rmax[1] > e? rmax[1] : e;
		if (t->len > max) max = t->len;
	}
	rmax[0] = rmax[0] > 0? rmax[0] : 0;
	rmax[1] = rmax[1] < l_pac<<1? rmax[1] : l_pac<<1;
	if (rmax[0] < l_pac && l_pac < rmax[1]) { // crossing the forward-reverse boundary; then choose one side
		if (c->seeds[0].rbeg < l_pac) rmax[1] = l_pac; // this works because all seeds are guaranteed to be on the same strand
		else rmax[0] = l_pac;
	}
	// retrieve the reference sequence
	rseq = bns_fetch_seq(bns, pac, &rmax[0], c->seeds[0].rbeg, &rmax[1], &rid);
	assert(c->rid == rid);

	srt = malloc(c->n * 8);
	for (i = 0; i < c->n; ++i)
		srt[i] = (uint64_t)c->seeds[i].score<<32 | i;
	ks_introsort_64(c->n, srt);

	for (k = c->n - 1; k >= 0; --k) {
		mem_alnreg_t *a;
		s = &c->seeds[(uint32_t)srt[k]];

		for (i = 0; i < av->n; ++i) { // test whether extension has been made before
			mem_alnreg_t *p = &av->a[i];
			int64_t rd;
			int qd, w, max_gap;
			if (s->rbeg < p->rb || s->rbeg + s->len > p->re || s->qbeg < p->qb || s->qbeg + s->len > p->qe) continue; // not fully contained
			if (s->len - p->seedlen0 > .1 * l_query) continue; // this seed may give a better alignment
			// qd: distance ahead of the seed on query; rd: on reference
			qd = s->qbeg - p->qb; rd = s->rbeg - p->rb;
			max_gap = cal_max_gap(opt, qd < rd? qd : rd); // the maximal gap allowed in regions ahead of the seed
			w = max_gap < p->w? max_gap : p->w; // bounded by the band width
			if (qd - rd < w && rd - qd < w) break; // the seed is "around" a previous hit
			// similar to the previous four lines, but this time we look at the region behind
			qd = p->qe - (s->qbeg + s->len); rd = p->re - (s->rbeg + s->len);
			max_gap = cal_max_gap(opt, qd < rd? qd : rd);
			w = max_gap < p->w? max_gap : p->w;
			if (qd - rd < w && rd - qd < w) break;
		}
		if (i < av->n) { // the seed is (almost) contained in an existing alignment; further testing is needed to confirm it is not leading to a different aln
			if (bwa_verbose >= 4)
				printf("** Seed(%d) [%ld;%ld,%ld] is almost contained in an existing alignment [%d,%d) <=> [%ld,%ld)\n",
					   k, (long)s->len, (long)s->qbeg, (long)s->rbeg, av->a[i].qb, av->a[i].qe, (long)av->a[i].rb, (long)av->a[i].re);
			for (i = k + 1; i < c->n; ++i) { // check overlapping seeds in the same chain
				const mem_seed_t *t;
				if (srt[i] == 0) continue;
				t = &c->seeds[(uint32_t)srt[i]];
				if (t->len < s->len * .95) continue; // only check overlapping if t is long enough; TODO: more efficient by early stopping
				if (s->qbeg <= t->qbeg && s->qbeg + s->len - t->qbeg >= s->len>>2 && t->qbeg - s->qbeg != t->rbeg - s->rbeg) break;
				if (t->qbeg <= s->qbeg && t->qbeg + t->len - s->qbeg >= s->len>>2 && s->qbeg - t->qbeg != s->rbeg - t->rbeg) break;
			}
			if (i == c->n) { // no overlapping seeds; then skip extension
				srt[k] = 0; // mark that seed extension has not been performed
				continue;
			}
			if (bwa_verbose >= 4)
				printf("** Seed(%d) might lead to a different alignment even though it is contained. Extension will be performed.\n", k);
		}

		a = kv_pushp(mem_alnreg_t, *av);
		memset(a, 0, sizeof(mem_alnreg_t));
		a->w = aw[0] = aw[1] = opt->w;
		a->score = a->truesc = -1;
		a->rid = c->rid;

		if (bwa_verbose >= 4) err_printf("** ---> Extending from seed(%d) [%ld;%ld,%ld] @ %s <---\n", k, (long)s->len, (long)s->qbeg, (long)s->rbeg, bns->anns[c->rid].name);
		if (s->qbeg) { // left extension
			uint8_t *rs, *qs;
			int qle, tle, gtle, gscore;
			qs = malloc(s->qbeg);
			for (i = 0; i < s->qbeg; ++i) qs[i] = query[s->qbeg - 1 - i];
			tmp = s->rbeg - rmax[0];
			rs = malloc(tmp);
			for (i = 0; i < tmp; ++i) rs[i] = rseq[tmp - 1 - i];
			for (i = 0; i < MAX_BAND_TRY; ++i) {
				int prev = a->score;
				aw[0] = opt->w << i;
				if (bwa_verbose >= 4) {
					int j;
					printf("*** Left ref:   "); for (j = 0; j < tmp; ++j) putchar("ACGTN"[(int)rs[j]]); putchar('\n');
					printf("*** Left query: "); for (j = 0; j < s->qbeg; ++j) putchar("ACGTN"[(int)qs[j]]); putchar('\n');
				}
				a->score = ksw_extend2(s->qbeg, qs, tmp, rs, 5, opt->mat, opt->o_del, opt->e_del, opt->o_ins, opt->e_ins, aw[0], opt->pen_clip5, opt->zdrop, s->len * opt->a, &qle, &tle, &gtle, &gscore, &max_off[0]);
				if (bwa_verbose >= 4) { printf("*** Left extension: prev_score=%d; score=%d; bandwidth=%d; max_off_diagonal_dist=%d\n", prev, a->score, aw[0], max_off[0]); fflush(stdout); }
				if (a->score == prev || max_off[0] < (aw[0]>>1) + (aw[0]>>2)) break;
			}
			// check whether we prefer to reach the end of the query
			if (gscore <= 0 || gscore <= a->score - opt->pen_clip5) { // local extension
				a->qb = s->qbeg - qle, a->rb = s->rbeg - tle;
				a->truesc = a->score;
			} else { // to-end extension
				a->qb = 0, a->rb = s->rbeg - gtle;
				a->truesc = gscore;
			}
			free(qs); free(rs);
		} else a->score = a->truesc = s->len * opt->a, a->qb = 0, a->rb = s->rbeg;

		if (s->qbeg + s->len != l_query) { // right extension
			int qle, tle, qe, re, gtle, gscore, sc0 = a->score;
			qe = s->qbeg + s->len;
			re = s->rbeg + s->len - rmax[0];
			assert(re >= 0);
			for (i = 0; i < MAX_BAND_TRY; ++i) {
				int prev = a->score;
				aw[1] = opt->w << i;
				if (bwa_verbose >= 4) {
					int j;
					printf("*** Right ref:   "); for (j = 0; j < rmax[1] - rmax[0] - re; ++j) putchar("ACGTN"[(int)rseq[re+j]]); putchar('\n');
					printf("*** Right query: "); for (j = 0; j < l_query - qe; ++j) putchar("ACGTN"[(int)query[qe+j]]); putchar('\n');
				}
				a->score = ksw_extend2(l_query - qe, query + qe, rmax[1] - rmax[0] - re, rseq + re, 5, opt->mat, opt->o_del, opt->e_del, opt->o_ins, opt->e_ins, aw[1], opt->pen_clip3, opt->zdrop, sc0, &qle, &tle, &gtle, &gscore, &max_off[1]);
				if (bwa_verbose >= 4) { printf("*** Right extension: prev_score=%d; score=%d; bandwidth=%d; max_off_diagonal_dist=%d\n", prev, a->score, aw[1], max_off[1]); fflush(stdout); }
				if (a->score == prev || max_off[1] < (aw[1]>>1) + (aw[1]>>2)) break;
			}
			// similar to the above
			if (gscore <= 0 || gscore <= a->score - opt->pen_clip3) { // local extension
				a->qe = qe + qle, a->re = rmax[0] + re + tle;
				a->truesc += a->score - sc0;
			} else { // to-end extension
				a->qe = l_query, a->re = rmax[0] + re + gtle;
				a->truesc += gscore - sc0;
			}
		} else a->qe = l_query, a->re = s->rbeg + s->len;
		if (bwa_verbose >= 4) printf("*** Added alignment region: [%d,%d) <=> [%ld,%ld); score=%d; {left,right}_bandwidth={%d,%d}\n", a->qb, a->qe, (long)a->rb, (long)a->re, a->score, aw[0], aw[1]);

		// compute seedcov
		for (i = 0, a->seedcov = 0; i < c->n; ++i) {
			const mem_seed_t *t = &c->seeds[i];
			if (t->qbeg >= a->qb && t->qbeg + t->len <= a->qe && t->rbeg >= a->rb && t->rbeg + t->len <= a->re) // seed fully contained
				a->seedcov += t->len; // this is not very accurate, but for approx. mapQ, this is good enough
		}
		a->w = aw[0] > aw[1]? aw[0] : aw[1];
		a->seedlen0 = s->len;

		a->frac_rep = c->frac_rep;
	}
	free(srt); free(rseq);
}

/*****************************
 * Basic hit->SAM conversion *
 *****************************/

static inline int infer_bw(int l1, int l2, int score, int a, int q, int r)
{
	int w;
	if (l1 == l2 && l1 * a - score < (q + r - a)<<1) return 0; // to get equal alignment length, we need at least two gaps
	w = ((double)((l1 < l2? l1 : l2) * a - score - q) / r + 2.);
	if (w < abs(l1 - l2)) w = abs(l1 - l2);
	return w;
}

static inline int get_rlen(int n_cigar, const uint32_t *cigar)
{
	int k, l;
	for (k = l = 0; k < n_cigar; ++k) {
		int op = cigar[k]&0xf;
		if (op == 0 || op == 2)
			l += cigar[k]>>4;
	}
	return l;
}

static inline void add_cigar(const mem_opt_t *opt, mem_aln_t *p, kstring_t *str, int which)
{
	int i;
	if (p->n_cigar) { // aligned
		for (i = 0; i < p->n_cigar; ++i) {
			int c = p->cigar[i]&0xf;
			if (!(opt->flag&MEM_F_SOFTCLIP) && !p->is_alt && (c == 3 || c == 4))
				c = which? 4 : 3; // use hard clipping for supplementary alignments
			kputw(p->cigar[i]>>4, str); kputc("MIDSH"[c], str);
		}
	} else kputc('*', str); // having a coordinate but unaligned (e.g. when copy_mate is true)
}

void mem_aln2sam(const mem_opt_t *opt, const bntseq_t *bns, kstring_t *str, bseq1_t *s, int n, const mem_aln_t *list, int which, const mem_aln_t *m_)
{
	int i, l_name;
	mem_aln_t ptmp = list[which], *p = &ptmp, mtmp, *m = 0; // make a copy of the alignment to convert

	if (m_) mtmp = *m_, m = &mtmp;
	// set flag
	p->flag |= m? 0x1 : 0; // is paired in sequencing
	p->flag |= p->rid < 0? 0x4 : 0; // is mapped
	p->flag |= m && m->rid < 0? 0x8 : 0; // is mate mapped
	if (p->rid < 0 && m && m->rid >= 0) // copy mate to alignment
		p->rid = m->rid, p->pos = m->pos, p->is_rev = m->is_rev, p->n_cigar = 0;
	if (m && m->rid < 0 && p->rid >= 0) // copy alignment to mate
		m->rid = p->rid, m->pos = p->pos, m->is_rev = p->is_rev, m->n_cigar = 0;
	p->flag |= p->is_rev? 0x10 : 0; // is on the reverse strand
	p->flag |= m && m->is_rev? 0x20 : 0; // is mate on the reverse strand

	// print up to CIGAR
	l_name = strlen(s->name);
	ks_resize(str, str->l + s->l_seq + l_name + (s->qual? s->l_seq : 0) + 20);
	kputsn(s->name, l_name, str); kputc('\t', str); // QNAME
	kputw((p->flag&0xffff) | (p->flag&0x10000? 0x100 : 0), str); kputc('\t', str); // FLAG
	if (p->rid >= 0) { // with coordinate
		kputs(bns->anns[p->rid].name, str); kputc('\t', str); // RNAME
		kputl(p->pos + 1, str); kputc('\t', str); // POS
		kputw(p->mapq, str); kputc('\t', str); // MAPQ
		add_cigar(opt, p, str, which);
	} else kputsn("*\t0\t0\t*", 7, str); // without coordinte
	kputc('\t', str);

	// print the mate position if applicable
	if (m && m->rid >= 0) {
		if (p->rid == m->rid) kputc('=', str);
		else kputs(bns->anns[m->rid].name, str);
		kputc('\t', str);
		kputl(m->pos + 1, str); kputc('\t', str);
		if (p->rid == m->rid) {
			int64_t p0 = p->pos + (p->is_rev? get_rlen(p->n_cigar, p->cigar) - 1 : 0);
			int64_t p1 = m->pos + (m->is_rev? get_rlen(m->n_cigar, m->cigar) - 1 : 0);
			if (m->n_cigar == 0 || p->n_cigar == 0) kputc('0', str);
			else kputl(-(p0 - p1 + (p0 > p1? 1 : p0 < p1? -1 : 0)), str);
		} else kputc('0', str);
	} else kputsn("*\t0\t0", 5, str);
	kputc('\t', str);

	// print SEQ and QUAL
	if (p->flag & 0x100) { // for secondary alignments, don't write SEQ and QUAL
		kputsn("*\t*", 3, str);
	} else if (!p->is_rev) { // the forward strand
		int i, qb = 0, qe = s->l_seq;
		if (p->n_cigar && which && !(opt->flag&MEM_F_SOFTCLIP) && !p->is_alt) { // have cigar && not the primary alignment && not softclip all
			if ((p->cigar[0]&0xf) == 4 || (p->cigar[0]&0xf) == 3) qb += p->cigar[0]>>4;
			if ((p->cigar[p->n_cigar-1]&0xf) == 4 || (p->cigar[p->n_cigar-1]&0xf) == 3) qe -= p->cigar[p->n_cigar-1]>>4;
		}
		ks_resize(str, str->l + (qe - qb) + 1);
		for (i = qb; i < qe; ++i) str->s[str->l++] = "ACGTN"[(int)s->seq[i]];
		kputc('\t', str);
		if (s->qual) { // printf qual
			ks_resize(str, str->l + (qe - qb) + 1);
			for (i = qb; i < qe; ++i) str->s[str->l++] = s->qual[i];
			str->s[str->l] = 0;
		} else kputc('*', str);
	} else { // the reverse strand
		int i, qb = 0, qe = s->l_seq;
		if (p->n_cigar && which && !(opt->flag&MEM_F_SOFTCLIP) && !p->is_alt) {
			if ((p->cigar[0]&0xf) == 4 || (p->cigar[0]&0xf) == 3) qe -= p->cigar[0]>>4;
			if ((p->cigar[p->n_cigar-1]&0xf) == 4 || (p->cigar[p->n_cigar-1]&0xf) == 3) qb += p->cigar[p->n_cigar-1]>>4;
		}
		ks_resize(str, str->l + (qe - qb) + 1);
		for (i = qe-1; i >= qb; --i) str->s[str->l++] = "TGCAN"[(int)s->seq[i]];
		kputc('\t', str);
		if (s->qual) { // printf qual
			ks_resize(str, str->l + (qe - qb) + 1);
			for (i = qe-1; i >= qb; --i) str->s[str->l++] = s->qual[i];
			str->s[str->l] = 0;
		} else kputc('*', str);
	}

	// print optional tags
	if (p->n_cigar) {
		kputsn("\tNM:i:", 6, str); kputw(p->NM, str);
		kputsn("\tMD:Z:", 6, str); kputs((char*)(p->cigar + p->n_cigar), str);
	}
	if (m && m->n_cigar) { kputsn("\tMC:Z:", 6, str); add_cigar(opt, m, str, which); }
	if (p->score >= 0) { kputsn("\tAS:i:", 6, str); kputw(p->score, str); }
	if (p->sub >= 0) { kputsn("\tXS:i:", 6, str); kputw(p->sub, str); }
	if (bwa_rg_id[0]) { kputsn("\tRG:Z:", 6, str); kputs(bwa_rg_id, str); }
	if (!(p->flag & 0x100)) { // not multi-hit
		for (i = 0; i < n; ++i)
			if (i != which && !(list[i].flag&0x100)) break;
		if (i < n) { // there are other primary hits; output them
			kputsn("\tSA:Z:", 6, str);
			for (i = 0; i < n; ++i) {
				const mem_aln_t *r = &list[i];
				int k;
				if (i == which || (r->flag&0x100)) continue; // proceed if: 1) different from the current; 2) not shadowed multi hit
				kputs(bns->anns[r->rid].name, str); kputc(',', str);
				kputl(r->pos+1, str); kputc(',', str);
				kputc("+-"[r->is_rev], str); kputc(',', str);
				for (k = 0; k < r->n_cigar; ++k) {
					kputw(r->cigar[k]>>4, str); kputc("MIDSH"[r->cigar[k]&0xf], str);
				}
				kputc(',', str); kputw(r->mapq, str);
				kputc(',', str); kputw(r->NM, str);
				kputc(';', str);
			}
		}
		if (p->alt_sc > 0)
			ksprintf(str, "\tpa:f:%.3f", (double)p->score / p->alt_sc);
	}
	if (p->XA) {
		kputsn((opt->flag&MEM_F_XB)? "\tXB:Z:" : "\tXA:Z:", 6, str);
		kputs(p->XA, str);
	}
	if (s->comment) { kputc('\t', str); kputs(s->comment, str); }
	if ((opt->flag&MEM_F_REF_HDR) && p->rid >= 0 && bns->anns[p->rid].anno != 0 && bns->anns[p->rid].anno[0] != 0) {
		int tmp;
		kputsn("\tXR:Z:", 6, str);
		tmp = str->l;
		kputs(bns->anns[p->rid].anno, str);
		for (i = tmp; i < str->l; ++i) // replace TAB in the comment to SPACE
			if (str->s[i] == '\t') str->s[i] = ' ';
	}
	kputc('\n', str);
}

/************************
 * Integrated interface *
 ************************/

int mem_approx_mapq_se(const mem_opt_t *opt, const mem_alnreg_t *a)
{
	int mapq, l, sub = a->sub? a->sub : opt->min_seed_len * opt->a;
	double identity;
	sub = a->csub > sub? a->csub : sub;
	if (sub >= a->score) return 0;
	l = a->qe - a->qb > a->re - a->rb? a->qe - a->qb : a->re - a->rb;
	identity = 1. - (double)(l * opt->a - a->score) / (opt->a + opt->b) / l;
	if (a->score == 0) {
		mapq = 0;
	} else if (opt->mapQ_coef_len > 0) {
		double tmp;
		tmp = l < opt->mapQ_coef_len? 1. : opt->mapQ_coef_fac / log(l);
		tmp *= identity * identity;
		mapq = (int)(6.02 * (a->score - sub) / opt->a * tmp * tmp + .499);
	} else {
		mapq = (int)(MEM_MAPQ_COEF * (1. - (double)sub / a->score) * log(a->seedcov) + .499);
		mapq = identity < 0.95? (int)(mapq * identity * identity + .499) : mapq;
	}
	if (a->sub_n > 0) mapq -= (int)(4.343 * log(a->sub_n+1) + .499);
	if (mapq > 60) mapq = 60;
	if (mapq < 0) mapq = 0;
	mapq = (int)(mapq * (1. - a->frac_rep) + .499);
	return mapq;
}

void mem_reorder_primary5(int T, mem_alnreg_v *a)
{
	int k, n_pri = 0, left_st = INT_MAX, left_k = -1;
	mem_alnreg_t t;
	for (k = 0; k < a->n; ++k)
		if (a->a[k].secondary < 0 && !a->a[k].is_alt && a->a[k].score >= T) ++n_pri;
	if (n_pri <= 1) return; // only one alignment
	for (k = 0; k < a->n; ++k) {
		mem_alnreg_t *p = &a->a[k];
		if (p->secondary >= 0 || p->is_alt || p->score < T) continue;
		if (p->qb < left_st) left_st = p->qb, left_k = k;
	}
	assert(a->a[0].secondary < 0);
	if (left_k == 0) return; // no need to reorder
	t = a->a[0], a->a[0] = a->a[left_k], a->a[left_k] = t;
	for (k = 1; k < a->n; ++k) { // update secondary and secondary_all
		mem_alnreg_t *p = &a->a[k];
		if (p->secondary == 0) p->secondary = left_k;
		else if (p->secondary == left_k) p->secondary = 0;
		if (p->secondary_all == 0) p->secondary_all = left_k;
		else if (p->secondary_all == left_k) p->secondary_all = 0;
	}
}

// TODO (future plan): group hits into a uint64_t[] array. This will be cleaner and more flexible
void mem_reg2sam(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, bseq1_t *s, mem_alnreg_v *a, int extra_flag, const mem_aln_t *m)
{
	extern char **mem_gen_alt(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, mem_alnreg_v *a, int l_query, const char *query);
	kstring_t str;
	kvec_t(mem_aln_t) aa;
	int k, l;
	char **XA = 0;

	if (!(opt->flag & MEM_F_ALL))
		XA = mem_gen_alt(opt, bns, pac, a, s->l_seq, s->seq);
	kv_init(aa);
	str.l = str.m = 0; str.s = 0;
	for (k = l = 0; k < a->n; ++k) {
		mem_alnreg_t *p = &a->a[k];
		mem_aln_t *q;
		if (p->score < opt->T) continue;
		if (p->secondary >= 0 && (p->is_alt || !(opt->flag&MEM_F_ALL))) continue;
		if (p->secondary >= 0 && p->secondary < INT_MAX && p->score < a->a[p->secondary].score * opt->drop_ratio) continue;
		q = kv_pushp(mem_aln_t, aa);
		*q = mem_reg2aln(opt, bns, pac, s->l_seq, s->seq, p);
		assert(q->rid >= 0); // this should not happen with the new code
		q->XA = XA? XA[k] : 0;
		q->flag |= extra_flag; // flag secondary
		if (p->secondary >= 0) q->sub = -1; // don't output sub-optimal score
		if (l && p->secondary < 0) // if supplementary
			q->flag |= (opt->flag&MEM_F_NO_MULTI)? 0x10000 : 0x800;
		if (!(opt->flag & MEM_F_KEEP_SUPP_MAPQ) && l && !p->is_alt && q->mapq > aa.a[0].mapq)
			q->mapq = aa.a[0].mapq; // lower mapq for supplementary mappings, unless -5 or -q is applied
		++l;
	}
	if (aa.n == 0) { // no alignments good enough; then write an unaligned record
		mem_aln_t t;
		t = mem_reg2aln(opt, bns, pac, s->l_seq, s->seq, 0);
		t.flag |= extra_flag;
		mem_aln2sam(opt, bns, &str, s, 1, &t, 0, m);
	} else {
		for (k = 0; k < aa.n; ++k)
			mem_aln2sam(opt, bns, &str, s, aa.n, aa.a, k, m);
		for (k = 0; k < aa.n; ++k) free(aa.a[k].cigar);
		free(aa.a);
	}
	s->sam = str.s;
	if (XA) {
		for (k = 0; k < a->n; ++k) free(XA[k]);
		free(XA);
	}
}

mem_alnreg_v mem_align1_core(const mem_opt_t *opt, const bwt_t *bwt, const bntseq_t *bns, const uint8_t *pac, int l_seq, char *seq, void *buf)
{
	int i;
	mem_chain_v chn;
	mem_alnreg_v regs;

	for (i = 0; i < l_seq; ++i) // convert to 2-bit encoding if we have not done so
		seq[i] = seq[i] < 4? seq[i] : nst_nt4_table[(int)seq[i]];

	chn = mem_chain(opt, bwt, bns, l_seq, (uint8_t*)seq, buf);
	chn.n = mem_chain_flt(opt, chn.n, chn.a);
	mem_flt_chained_seeds(opt, bns, pac, l_seq, (uint8_t*)seq, chn.n, chn.a);
	if (bwa_verbose >= 4) mem_print_chain(bns, &chn);

	kv_init(regs);
	for (i = 0; i < chn.n; ++i) {
		mem_chain_t *p = &chn.a[i];
		if (bwa_verbose >= 4) err_printf("* ---> Processing chain(%d) <---\n", i);
		mem_chain2aln(opt, bns, pac, l_seq, (uint8_t*)seq, p, &regs);
		free(chn.a[i].seeds);
	}
	free(chn.a);
	regs.n = mem_sort_dedup_patch(opt, bns, pac, (uint8_t*)seq, regs.n, regs.a);
	if (bwa_verbose >= 4) {
		err_printf("* %ld chains remain after removing duplicated chains\n", regs.n);
		for (i = 0; i < regs.n; ++i) {
			mem_alnreg_t *p = &regs.a[i];
			printf("** %d, [%d,%d) <=> [%ld,%ld)\n", p->score, p->qb, p->qe, (long)p->rb, (long)p->re);
		}
	}
	for (i = 0; i < regs.n; ++i) {
		mem_alnreg_t *p = &regs.a[i];
		if (p->rid >= 0 && bns->anns[p->rid].is_alt)
			p->is_alt = 1;
	}
	return regs;
}

mem_aln_t mem_reg2aln(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, int l_query, const char *query_, const mem_alnreg_t *ar)
{
	mem_aln_t a;
	int i, w2, tmp, qb, qe, NM, score, is_rev, last_sc = -(1<<30), l_MD;
	int64_t pos, rb, re;
	uint8_t *query;

	memset(&a, 0, sizeof(mem_aln_t));
	if (ar == 0 || ar->rb < 0 || ar->re < 0) { // generate an unmapped record
		a.rid = -1; a.pos = -1; a.flag |= 0x4;
		return a;
	}
	qb = ar->qb, qe = ar->qe;
	rb = ar->rb, re = ar->re;
	query = malloc(l_query);
	for (i = 0; i < l_query; ++i) // convert to the nt4 encoding
		query[i] = query_[i] < 5? query_[i] : nst_nt4_table[(int)query_[i]];
	a.mapq = ar->secondary < 0? mem_approx_mapq_se(opt, ar) : 0;
	if (ar->secondary >= 0) a.flag |= 0x100; // secondary alignment
	tmp = infer_bw(qe - qb, re - rb, ar->truesc, opt->a, opt->o_del, opt->e_del);
	w2  = infer_bw(qe - qb, re - rb, ar->truesc, opt->a, opt->o_ins, opt->e_ins);
	w2 = w2 > tmp? w2 : tmp;
	if (bwa_verbose >= 4) printf("* Band width: inferred=%d, cmd_opt=%d, alnreg=%d\n", w2, opt->w, ar->w);
	if (w2 > opt->w) w2 = w2 < ar->w? w2 : ar->w;
	i = 0; a.cigar = 0;
	do {
		free(a.cigar);
		w2 = w2 < opt->w<<2? w2 : opt->w<<2;
		a.cigar = bwa_gen_cigar2(opt->mat, opt->o_del, opt->e_del, opt->o_ins, opt->e_ins, w2, bns->l_pac, pac, qe - qb, (uint8_t*)&query[qb], rb, re, &score, &a.n_cigar, &NM);
		if (bwa_verbose >= 4) printf("* Final alignment: w2=%d, global_sc=%d, local_sc=%d\n", w2, score, ar->truesc);
		if (score == last_sc || w2 == opt->w<<2) break; // it is possible that global alignment and local alignment give different scores
		last_sc = score;
		w2 <<= 1;
	} while (++i < 3 && score < ar->truesc - opt->a);
	l_MD = strlen((char*)(a.cigar + a.n_cigar)) + 1;
	a.NM = NM;
	pos = bns_depos(bns, rb < bns->l_pac? rb : re - 1, &is_rev);
	a.is_rev = is_rev;
	if (a.n_cigar > 0) { // squeeze out leading or trailing deletions
		if ((a.cigar[0]&0xf) == 2) {
			pos += a.cigar[0]>>4;
			--a.n_cigar;
			memmove(a.cigar, a.cigar + 1, a.n_cigar * 4 + l_MD);
		} else if ((a.cigar[a.n_cigar-1]&0xf) == 2) {
			--a.n_cigar;
			memmove(a.cigar + a.n_cigar, a.cigar + a.n_cigar + 1, l_MD); // MD needs to be moved accordingly
		}
	}
	if (qb != 0 || qe != l_query) { // add clipping to CIGAR
		int clip5, clip3;
		clip5 = is_rev? l_query - qe : qb;
		clip3 = is_rev? qb : l_query - qe;
		a.cigar = realloc(a.cigar, 4 * (a.n_cigar + 2) + l_MD);
		if (clip5) {
			memmove(a.cigar+1, a.cigar, a.n_cigar * 4 + l_MD); // make room for 5'-end clipping
			a.cigar[0] = clip5<<4 | 3;
			++a.n_cigar;
		}
		if (clip3) {
			memmove(a.cigar + a.n_cigar + 1, a.cigar + a.n_cigar, l_MD); // make room for 3'-end clipping
			a.cigar[a.n_cigar++] = clip3<<4 | 3;
		}
	}
	a.rid = bns_pos2rid(bns, pos);
	assert(a.rid == ar->rid);
	a.pos = pos - bns->anns[a.rid].offset;
	a.score = ar->score; a.sub = ar->sub > ar->csub? ar->sub : ar->csub;
	a.is_alt = ar->is_alt; a.alt_sc = ar->alt_sc;
	free(query);
	return a;
}

typedef struct {
	const mem_opt_t *opt;
	const bwt_t *bwt;
	const bntseq_t *bns;
	const uint8_t *pac;
	const mem_pestat_t *pes;
	smem_aux_t **aux;
	bseq1_t *seqs;
	mem_alnreg_v *regs;
	int64_t n_processed;
} worker_t;

static void worker1(void *data, int i, int tid)
{
	worker_t *w = (worker_t*)data;
	if (!(w->opt->flag&MEM_F_PE)) {
		if (bwa_verbose >= 4) printf("=====> Processing read '%s' <=====\n", w->seqs[i].name);
		w->regs[i] = mem_align1_core(w->opt, w->bwt, w->bns, w->pac, w->seqs[i].l_seq, w->seqs[i].seq, w->aux[tid]);
	} else {
		if (bwa_verbose >= 4) printf("=====> Processing read '%s'/1 <=====\n", w->seqs[i<<1|0].name);
		w->regs[i<<1|0] = mem_align1_core(w->opt, w->bwt, w->bns, w->pac, w->seqs[i<<1|0].l_seq, w->seqs[i<<1|0].seq, w->aux[tid]);
		if (bwa_verbose >= 4) printf("=====> Processing read '%s'/2 <=====\n", w->seqs[i<<1|1].name);
		w->regs[i<<1|1] = mem_align1_core(w->opt, w->bwt, w->bns, w->pac, w->seqs[i<<1|1].l_seq, w->seqs[i<<1|1].seq, w->aux[tid]);
	}
}

static void worker2(void *data, int i, int tid)
{
	extern int mem_sam_pe(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, const mem_pestat_t pes[4], uint64_t id, bseq1_t s[2], mem_alnreg_v a[2]);
	extern void mem_reg2ovlp(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, bseq1_t *s, mem_alnreg_v *a);
	worker_t *w = (worker_t*)data;
	if (!(w->opt->flag&MEM_F_PE)) {
		if (bwa_verbose >= 4) printf("=====> Finalizing read '%s' <=====\n", w->seqs[i].name);
		mem_mark_primary_se(w->opt, w->regs[i].n, w->regs[i].a, w->n_processed + i);
		if (w->opt->flag & MEM_F_PRIMARY5) mem_reorder_primary5(w->opt->T, &w->regs[i]);
		mem_reg2sam(w->opt, w->bns, w->pac, &w->seqs[i], &w->regs[i], 0, 0);
		free(w->regs[i].a);
	} else {
		if (bwa_verbose >= 4) printf("=====> Finalizing read pair '%s' <=====\n", w->seqs[i<<1|0].name);
		mem_sam_pe(w->opt, w->bns, w->pac, w->pes, (w->n_processed>>1) + i, &w->seqs[i<<1], &w->regs[i<<1]);
		free(w->regs[i<<1|0].a); free(w->regs[i<<1|1].a);
	}
}

void mem_process_seqs(const mem_opt_t *opt, const bwt_t *bwt, const bntseq_t *bns, const uint8_t *pac, int64_t n_processed, int n, bseq1_t *seqs, const mem_pestat_t *pes0)
{
	extern void kt_for(int n_threads, void (*func)(void*,int,int), void *data, int n);
	worker_t w;
	mem_pestat_t pes[4];
	double ctime, rtime;
	int i;

	ctime = cputime(); rtime = realtime();
	global_bns = bns;
	w.regs = malloc(n * sizeof(mem_alnreg_v));
	w.opt = opt; w.bwt = bwt; w.bns = bns; w.pac = pac;
	w.seqs = seqs; w.n_processed = n_processed;
	w.pes = &pes[0];
	w.aux = malloc(opt->n_threads * sizeof(smem_aux_t));
	for (i = 0; i < opt->n_threads; ++i)
		w.aux[i] = smem_aux_init();
	kt_for(opt->n_threads, worker1, &w, (opt->flag&MEM_F_PE)? n>>1 : n); // find mapping positions
	for (i = 0; i < opt->n_threads; ++i)
		smem_aux_destroy(w.aux[i]);
	free(w.aux);
	if (opt->flag&MEM_F_PE) { // infer insert sizes if not provided
		if (pes0) memcpy(pes, pes0, 4 * sizeof(mem_pestat_t)); // if pes0 != NULL, set the insert-size distribution as pes0
		else mem_pestat(opt, bns->l_pac, n, w.regs, pes); // otherwise, infer the insert size distribution from data
	}
	kt_for(opt->n_threads, worker2, &w, (opt->flag&MEM_F_PE)? n>>1 : n); // generate alignment
	free(w.regs);
	if (bwa_verbose >= 3)
		fprintf(stderr, "[M::%s] Processed %d reads in %.3f CPU sec, %.3f real sec\n", __func__, n, cputime() - ctime, realtime() - rtime);
}