collect contig alignment + renew_utg bug fixed

Makefile

@@ -6,7 +6,7 @@ CPPFLAGS=
 INCLUDES=
 OBJS=		CommandLines.o Process_Read.o Assembly.o Hash_Table.o \
 			POA.o Correct.o Levenshtein_distance.o Overlaps.o Trio.o kthread.o Purge_Dups.o \
-			htab.o hist.o sketch.o anchor.o extract.o sys.o ksw2_extz2_sse.o hic.o rcut.o
+			htab.o hist.o sketch.o anchor.o extract.o sys.o ksw2_extz2_sse.o hic.o rcut.o horder.o
 EXE=		hifiasm
 LIBS=		-lz -lpthread -lm
 
@@ -73,3 +73,4 @@ sketch.o: kvec.h htab.h Process_Read.h Overlaps.h kdq.h CommandLines.h
 sys.o: htab.h Process_Read.h Overlaps.h kvec.h kdq.h CommandLines.h
 hic.o: hic.h
 rcut.o: rcut.h
+horder.o: horder.h

Overlaps.cpp

@@ -12720,6 +12720,21 @@ R_to_U* ruIndex, float chimeric_rate, float drop_ratio, int max_hang, int min_ov
 long long gap_fuzz, bub_label_t* b_mask_t)
 { 
     hic_clean(sg);
+    ug_opt_t opt; memset(&opt, 0, sizeof(opt));
+    opt.coverage_cut = coverage_cut;
+    opt.sources = sources;
+    opt.reverse_sources = reverse_sources; 
+    opt.tipsLen = (asm_opt.max_short_tip*2);
+    opt.tip_drop_ratio = 0.15;
+    opt.stops_threshold = 3;
+    opt.ruIndex = ruIndex; 
+    opt.chimeric_rate = 0.05;
+    opt.drop_ratio = 0.9;
+    opt.max_hang = max_hang;
+    opt.min_ovlp = min_ovlp;
+    opt.is_bench = 0;
+    opt.b_mask_t = b_mask_t;
+
 
     kvec_asg_arc_t_warp new_rtg_edges, d_edges;
     kv_init(new_rtg_edges.a); kv_init(d_edges.a);
@@ -12773,7 +12788,7 @@ long long gap_fuzz, bub_label_t* b_mask_t)
     fclose(output_file);
     free(gfa_name);
 
-    hic_analysis(ug, sg, cov?cov->t_ch:t_ch);
+    hic_analysis(ug, sg, cov?cov->t_ch:t_ch, &opt);
 
     if(cov) destory_hap_cov_t(&cov);
     if(t_ch) destory_trans_chain(&t_ch);
@@ -13907,6 +13922,7 @@ void renew_utg(ma_ug_t **ug, asg_t* read_g, kvec_asg_arc_t_warp* edge)
         u = &(high_level_ug->u.a[i]);
         if(u->m == 0) continue;
         merge_unitig_content(u, (*ug), read_g, edge);
+        high_level_ug->g->seq[i].len = u->len;
     }
     ma_ug_destroy((*ug));
     (*ug) = high_level_ug;
@@ -16115,6 +16131,8 @@ kvec_asg_arc_t_warp* new_rtg_edges, bub_label_t* b_mask_t)
 
     set_drop_trio_flag(*ug);
     destory_hap_cov_t(&cov);
+
+    renew_utg(ug, read_g, new_rtg_edges);
 }
 
 

Overlaps.h

@@ -1195,6 +1195,28 @@ uint32_t get_u_trans_spec(kv_u_trans_t *ta, uint32_t qn, uint32_t tn, u_trans_t
 int ma_ug_seq(ma_ug_t *g, asg_t *read_g, ma_sub_t *coverage_cut, ma_hit_t_alloc* sources, 
 kvec_asg_arc_t_warp* edge, int max_hang, int min_ovlp, kvec_asg_arc_t_warp *E, uint32_t is_polish);
 
+
+typedef struct{
+    ma_sub_t* coverage_cut;
+    ma_hit_t_alloc* sources; 
+    ma_hit_t_alloc* reverse_sources; 
+    long long tipsLen; 
+    float tip_drop_ratio; 
+    long long stops_threshold; 
+    R_to_U* ruIndex; 
+    float chimeric_rate;
+    float drop_ratio; 
+    int max_hang;
+    int min_ovlp;
+    int is_bench;
+    bub_label_t* b_mask_t;
+}ug_opt_t;
+
+void adjust_utg_by_trio(ma_ug_t **ug, asg_t* read_g, uint8_t flag, float drop_rate,
+ma_hit_t_alloc* sources, ma_hit_t_alloc* reverse_sources, ma_sub_t* coverage_cut, 
+long long tipsLen, float tip_drop_ratio, long long stops_threshold, 
+R_to_U* ruIndex, float chimeric_rate, float drop_ratio, int max_hang, int min_ovlp,
+kvec_asg_arc_t_warp* new_rtg_edges, bub_label_t* b_mask_t);
 #define JUNK_COV 5
 #define DISCARD_RATE 0.8
 

hic.cpp

@@ -14,6 +14,7 @@
 #include "ksort.h"
 #include "kseq.h" // FASTA/Q parser
 #include "kdq.h"
+#include "horder.h"
 KSEQ_INIT(gzFile, gzread)
 KDQ_INIT(uint64_t)
 
@@ -190,22 +191,10 @@ typedef struct {
     uint64_t n_u;
 } kvec_pe_hit_hap;
 
-
-typedef struct {
-	uint64_t s, e, id, len;
-} pe_hit;
-
-typedef struct {
-    kvec_t(pe_hit) a;
-    kvec_t(uint64_t) idx;
-    kvec_t(uint64_t) occ;
-} kvec_pe_hit;
-
 typedef struct {
     kvec_t(hc_edge) a;
 }kvec_hc_edge;
 
-
 #define pe_hit_an1_key(x) ((x).s)
 KRADIX_SORT_INIT(pe_hit_an1, pe_hit, pe_hit_an1_key, member_size(pe_hit, s))
 #define pe_hit_an2_key(x) ((x).e)
@@ -286,6 +275,11 @@ KRADIX_SORT_INIT(hc_s_hit_off_cnt, s_hit, hc_s_hit_off_cnt_key, 8)
 KRADIX_SORT_INIT(hc_edge_u, hc_edge, hc_edge_key_u, 4)
 #define hc_edge_key_d(a) ((a).dis)
 KRADIX_SORT_INIT(hc_edge_d, hc_edge, hc_edge_key_d, member_size(hc_edge, dis))
+#define get_hit_suid(x, k) (((x).a.a[(k)].s<<1)>>(64 - (x).uID_bits))
+#define get_hit_spos(x, k) ((x).a.a[(k)].s & (x).pos_mode)
+#define get_hit_euid(x, k) (((x).a.a[(k)].e<<1)>>(64 - (x).uID_bits))
+#define get_hit_epos(x, k) ((x).a.a[(k)].e & (x).pos_mode)
+
 
 typedef struct {
     kvec_t(kvec_t_u64_warp) matrix;
@@ -302,7 +296,6 @@ typedef struct{
     uint64_t max, med;
 } trans_idx;
 
-
 reads_t R1, R2;
 ha_ug_index* ug_index;
 
@@ -15431,7 +15424,7 @@ void print_kv_u_trans_t(kv_u_trans_t *ta)
     fprintf(stderr, "[M::%s::] \n", __func__);
 }
 
-int hic_short_align(const enzyme *fn1, const enzyme *fn2, ha_ug_index* idx)
+int hic_short_align(const enzyme *fn1, const enzyme *fn2, ha_ug_index* idx, ug_opt_t *opt)
 {
     double index_time = yak_realtime();
     sldat_t sl;
@@ -15503,6 +15496,8 @@ int hic_short_align(const enzyme *fn1, const enzyme *fn2, ha_ug_index* idx)
 
     verbose_het_stat(&bub);
 
+    horder_t *ho = init_horder_t(&sl.hits, idx->uID_bits, idx->pos_mode, idx->read_g, idx->ug, &bub, opt);
+
     ///print_hc_links(&link, 0, &hap);
     // print_kv_u_trans(&k_trans, &link, s->s);
 
@@ -15514,6 +15509,7 @@ int hic_short_align(const enzyme *fn1, const enzyme *fn2, ha_ug_index* idx)
     ///print_debug_bubble_graph(&bub, idx->ug, asm_opt.output_file_name);
     // print_bubble_chain(&bub);
     // destory_contig_partition(&hap);
+    destory_horder_t(&ho);
     kv_destroy(back_hc_edge.a);
     kv_destroy(sl.hits.a);
     kv_destroy(sl.hits.idx);
@@ -15542,7 +15538,7 @@ int hic_short_align(const enzyme *fn1, const enzyme *fn2, ha_ug_index* idx)
 }
 
 
-void hic_analysis(ma_ug_t *ug, asg_t* read_g, trans_chain* t_ch)
+void hic_analysis(ma_ug_t *ug, asg_t* read_g, trans_chain* t_ch, ug_opt_t *opt)
 {
     ug_index = NULL;
     int exist = (asm_opt.load_index_from_disk? 
@@ -15553,7 +15549,7 @@ void hic_analysis(ma_ug_t *ug, asg_t* read_g, trans_chain* t_ch)
     ug_index->read_g = read_g;
     ug_index->t_ch = t_ch;
     ///test_unitig_index(ug_index, ug);
-    hic_short_align(asm_opt.hic_reads[0], asm_opt.hic_reads[1], ug_index);
+    hic_short_align(asm_opt.hic_reads[0], asm_opt.hic_reads[1], ug_index, opt);
     
     destory_hc_pt_index(ug_index);
 }

hic.h

@@ -11,7 +11,7 @@
 
 hc_edge* get_hc_edge(hc_links* link, uint64_t src, uint64_t dest, uint64_t dir);
 hc_edge* push_hc_edge(hc_linkeage* x, uint64_t uID, double weight, int dir, uint64_t* d);
-void hic_analysis(ma_ug_t *ug, asg_t* read_g, trans_chain* t_ch);
+void hic_analysis(ma_ug_t *ug, asg_t* read_g, trans_chain* t_ch, ug_opt_t *opt);
 void hic_benchmark(ma_ug_t *ug, asg_t* read_g);
 
 typedef struct {
@@ -54,6 +54,19 @@ typedef struct {
     int8_t *s;
     uint64_t xs;
 } ps_t;
+
+typedef struct {
+	uint64_t s, e, id, len;
+} pe_hit;
+
+typedef struct {
+    kvec_t(pe_hit) a;
+    kvec_t(uint64_t) idx;
+    kvec_t(uint64_t) occ;
+    uint64_t uID_bits;
+    uint64_t pos_mode;
+} kvec_pe_hit;
+
 #define P_het(B) ((B).num.n)
 #define M_het(B) ((B).num.n + 1)
 // #define IF_BUB(ID, B) ((B).index[(ID)] < (B).num.n)

horder.cpp

@@ -0,0 +1,348 @@
+#define __STDC_LIMIT_MACROS
+#include "float.h"
+#include "horder.h"
+#include <math.h>
+#include "hic.h"
+#include "htab.h"
+#include "assert.h"
+#include "Overlaps.h"
+#include "Hash_Table.h"
+#include "Correct.h"
+#include "Purge_Dups.h"
+#include "rcut.h"
+#include "khashl.h"
+#include "kthread.h"
+#include "ksort.h"
+#include "kseq.h" // FASTA/Q parser
+#include "kdq.h"
+KSEQ_INIT(gzFile, gzread)
+KDQ_INIT(uint64_t)
+#define pe_hit_an1_idx_key(x) ((x).s<<1)
+KRADIX_SORT_INIT(pe_hit_idx_hn1, pe_hit, pe_hit_an1_idx_key, member_size(pe_hit, s))
+#define pe_hit_an2_idx_key(x) ((x).e<<1)
+KRADIX_SORT_INIT(pe_hit_idx_hn2, pe_hit, pe_hit_an2_idx_key, member_size(pe_hit, e))
+#define get_hit_suid(x, k) (((x).a.a[(k)].s<<1)>>(64 - (x).uID_bits))
+#define get_hit_spos(x, k) ((x).a.a[(k)].s & (x).pos_mode)
+#define get_hit_euid(x, k) (((x).a.a[(k)].e<<1)>>(64 - (x).uID_bits))
+#define get_hit_epos(x, k) ((x).a.a[(k)].e & (x).pos_mode)
+
+typedef struct {
+	uint64_t ruid;
+    uint64_t off;
+} hit_aux_t;
+
+typedef struct {
+	hit_aux_t *a;
+    size_t n, m;
+    kvec_t(uint64_t) idx;
+} u_hits_t;
+
+
+#define hit_aux_ruid_key(x) ((x).ruid)
+KRADIX_SORT_INIT(hit_aux_ruid, hit_aux_t, hit_aux_ruid_key, member_size(hit_aux_t, ruid))
+
+
+void get_r_hits(kvec_pe_hit *u_hits, kvec_pe_hit *r_hits, asg_t* r_g, ma_ug_t* ug, bubble_type* bub, uint64_t uID_bits, uint64_t pos_mode)
+{
+    uint64_t k, l, i, r_i, offset, rid, rev, rBeg, rEnd, ubits, p_mode, upos, rpos, update;
+    ma_utg_t *u = NULL;
+    memset(r_hits, 0, sizeof(*r_hits));
+    r_hits->uID_bits = uID_bits;
+    r_hits->pos_mode = pos_mode;
+    //reset for reads
+    for (ubits=1; (uint64_t)(1<<ubits)<(uint64_t)r_g->n_seq; ubits++);
+    p_mode = ((uint64_t)-1) >> (ubits + 1);
+
+    kv_malloc(r_hits->a, u_hits->a.n); r_hits->a.n = r_hits->a.m = u_hits->a.n;
+    memcpy(r_hits->a.a, u_hits->a.a, r_hits->a.n*sizeof(pe_hit));
+    radix_sort_pe_hit_idx_hn1(r_hits->a.a, r_hits->a.a + r_hits->a.n);
+    for (k = 1, l = 0; k <= r_hits->a.n; ++k) 
+    {   
+        if (k == r_hits->a.n || get_hit_suid(*r_hits, k) != get_hit_suid(*r_hits, l))//same suid
+        {
+            ///already sort by spos
+            u = &(ug->u.a[get_hit_suid(*r_hits, l)]);
+            update = 0;
+            for (i = offset = 0, r_i = l; i < u->n; i++)
+            {
+                rid = u->a[i]>>33; 
+                rBeg = offset;
+                rEnd = rBeg + r_g->seq[rid].len - 1;
+                for (; r_i < k; r_i++)
+                {
+                    upos = get_hit_spos(*r_hits, r_i);///pos at unitig
+                    if(upos > rEnd) break;
+                    if(upos >= rBeg && upos <= rEnd)
+                    {
+                        if(bub)
+                        {
+                            r_hits->a.a[r_i].id = (uint32_t)r_hits->a.a[r_i].id;
+                            if(!IF_HOM(get_hit_suid(*r_hits, r_i), *bub))
+                            {
+                                r_hits->a.a[r_i].id += ((uint64_t)(1)<<32);
+                            }
+                        }
+
+                        rpos = (((u->a[i]>>32)&1)? rEnd - upos : upos - rBeg);///pos at read
+                        rev = ((u->a[i]>>32)&1) ^ (r_hits->a.a[r_i].s>>63);
+                        r_hits->a.a[r_i].s = (rev<<63) | ((rid << (64-ubits))>>1) | (rpos & p_mode);
+                        
+                        update++;
+                    }
+                }
+                offset += (uint32_t)u->a[i];
+            }
+
+            if(r_i != k || update != k - l) fprintf(stderr, "ERROR-r_i\n");
+            l = k;
+        }
+    }
+
+    radix_sort_pe_hit_idx_hn2(r_hits->a.a, r_hits->a.a + r_hits->a.n);
+    for (k = 1, l = 0; k <= r_hits->a.n; ++k) 
+    {
+        if (k == r_hits->a.n || get_hit_euid(*r_hits, k) != get_hit_euid(*r_hits, l))//same euid
+        {
+            ///already sort by epos
+            u = &(ug->u.a[get_hit_euid(*r_hits, l)]);
+            update = 0;
+            for (i = offset = 0, r_i = l; i < u->n; i++)
+            {
+                rid = u->a[i]>>33; 
+                rBeg = offset;
+                rEnd = rBeg + r_g->seq[rid].len - 1;
+                for (; r_i < k; r_i++)
+                {
+                    upos = get_hit_epos(*r_hits, r_i);///pos at unitig
+                    if(upos > rEnd) break;
+                    if(upos >= rBeg && upos <= rEnd)
+                    {
+                        if(bub)
+                        {
+                            r_hits->a.a[r_i].id >>= 32;
+                            r_hits->a.a[r_i].id <<= 32;
+                            if(!IF_HOM(get_hit_euid(*r_hits, r_i), *bub))
+                            {
+                                r_hits->a.a[r_i].id += 1;
+                            }
+                        }
+
+                        rpos = (((u->a[i]>>32)&1)? rEnd - upos : upos - rBeg);///pos at read
+                        rev = ((u->a[i]>>32)&1) ^ (r_hits->a.a[r_i].e>>63);
+                        r_hits->a.a[r_i].e = (rev<<63) | ((rid << (64-ubits))>>1) | (rpos & p_mode);
+                        
+                        update++;
+                    }
+                }
+                offset += (uint32_t)u->a[i];
+            }
+
+            if(r_i != k || update != k - l) fprintf(stderr, "ERROR-r_i\n");
+            l = k;
+        }
+    }
+
+    r_hits->uID_bits = ubits;
+    r_hits->pos_mode = p_mode;
+}
+
+uint64_t get_corresp_usite(uint64_t rid, uint64_t rpos, uint64_t rev, uint64_t rlen, u_hits_t *x, uint64_t ubits, uint64_t p_mode, kvec_t_u64_warp *buf)
+{
+    hit_aux_t *a = NULL;
+    uint64_t a_n, i, new_uid, new_pos, new_rev;
+    
+    a = x->a + (x->idx.a[rid]>>32);
+    a_n = (uint32_t)x->idx.a[rid];
+    for (i = 0; i < a_n; i++)
+    {
+        new_uid = (uint32_t)a[i].ruid;
+        new_pos = (((a[i].ruid>>32)&1)? a[i].off + rlen - 1 - rpos : a[i].off + rpos);
+        new_rev = ((a[i].ruid>>32)&1)^rev;
+        kv_push(uint64_t, buf->a, (new_rev<<63) | ((new_uid << (64-ubits))>>1) | (new_pos & p_mode));
+    }
+
+    return a_n;
+}
+
+void update_u_hits(kvec_pe_hit *u_hits, kvec_pe_hit *r_hits, ma_ug_t* ug, asg_t* r_g)
+{
+    u_hits_t x; memset(&x, 0, sizeof(x));
+    hit_aux_t *p = NULL;
+    ma_utg_t *u = NULL;
+    pe_hit *t = NULL;
+    uint64_t v, i, l, k, offset, occ_1, occ_2, *a_1, *a_2, i_1, i_2;
+    
+    for (v = 0; v < ug->g->n_seq; v++)
+    {
+        u = &(ug->u.a[v]);
+        for (i = offset = 0; i < u->n; i++)
+        {
+            kv_pushp(hit_aux_t, x, &p);
+            p->ruid = u->a[i]>>32; 
+            p->ruid <<= 32;
+            p->ruid |= v;
+            p->off = offset;
+            offset += (uint32_t)u->a[i];
+        }
+    }
+
+    radix_sort_hit_aux_ruid(x.a, x.a + x.n);
+    x.idx.n = x.idx.m = (x.n?(x.a[x.n-1].ruid>>33)+1:0);
+    CALLOC(x.idx.a, x.idx.n);
+    for (k = 1, l = 0; k <= x.n; ++k) 
+    {   
+        if (k == x.n || (x.a[k].ruid>>33) != (x.a[l].ruid>>33))//same rid
+        {
+            x.idx.a[x.a[l].ruid>>33] = (uint64_t)l << 32 | (k - l);
+            l = k;
+        }
+    }
+
+    u_hits->a.n = u_hits->idx.n = u_hits->occ.n = 0;
+    for (u_hits->uID_bits=1; (uint64_t)(1<<u_hits->uID_bits)<(uint64_t)ug->g->n_seq; u_hits->uID_bits++);
+    u_hits->pos_mode = ((uint64_t)-1) >> (u_hits->uID_bits + 1);
+    kvec_t_u64_warp buf; kv_init(buf.a);
+
+    for (i = 0; i < r_hits->a.n; i++)
+    {
+        buf.a.n = 0;
+        occ_1 = get_corresp_usite(get_hit_suid(*r_hits, i), get_hit_spos(*r_hits, i), 
+        r_hits->a.a[i].s>>63, r_g->seq[get_hit_suid(*r_hits, i)].len, &x, u_hits->uID_bits, 
+        u_hits->pos_mode, &buf);
+        occ_2 = get_corresp_usite(get_hit_euid(*r_hits, i), get_hit_epos(*r_hits, i),
+        r_hits->a.a[i].e>>63, r_g->seq[get_hit_euid(*r_hits, i)].len, &x, u_hits->uID_bits, 
+        u_hits->pos_mode, &buf);
+        if(occ_1 == 0 || occ_2 == 0) continue;
+        a_1 = buf.a.a; a_2 = buf.a.a + occ_1;
+        for (i_1 = 0; i_1 < occ_1; i_1++)
+        {
+            for (i_2 = 0; i_2 < occ_2; i_2++)
+            {
+                kv_pushp(pe_hit, u_hits->a, &t);
+                t->id = r_hits->a.a[i].id;
+                t->len = r_hits->a.a[i].len;
+                t->s = a_1[i_1];
+                t->e = a_2[i_2];
+            }
+        }
+    }
+    
+
+    free(x.a); free(x.idx.a); kv_destroy(buf.a);
+}
+
+ma_ug_t* get_trio_unitig_graph(asg_t *sg, uint8_t flag, ug_opt_t *opt)
+{
+    kvec_asg_arc_t_warp new_rtg_edges;
+    kv_init(new_rtg_edges.a);
+
+    ma_ug_t *ug = NULL;
+    ug = ma_ug_gen(sg);
+
+    adjust_utg_by_trio(&ug, sg, flag, TRIO_THRES, opt->sources, opt->reverse_sources, 
+    opt->coverage_cut, opt->tipsLen, opt->tip_drop_ratio, opt->stops_threshold, 
+    opt->ruIndex, opt->chimeric_rate, opt->drop_ratio, opt->max_hang, opt->min_ovlp,
+    &new_rtg_edges, opt->b_mask_t);
+
+    kv_destroy(new_rtg_edges.a);
+    return ug;
+}
+
+void generate_haplotypes(horder_t *h, ug_opt_t *opt)
+{
+    uint64_t i, off;
+    ma_ug_t *ug_1 = NULL, *ug_2 = NULL;
+    asg_cleanup(h->r_g);
+    asg_arc_del_trans(h->r_g, asm_opt.gap_fuzz);///must
+
+    ug_1 = get_trio_unitig_graph(h->r_g, FATHER, opt);
+    ug_2 = get_trio_unitig_graph(h->r_g, MOTHER, opt);
+    h->ug = ug_1; 
+
+    off = ug_1->g->n_seq;
+    ///update graph
+    for (i = 0; i < ug_2->g->n_seq; i++)
+    {
+        asg_seq_set(h->ug->g, off + i, ug_2->g->seq[i].len, ug_2->g->seq[i].del);
+    }
+    
+    asg_arc_t *p = NULL;
+    for (i = 0; i < ug_2->g->n_arc; i++)
+    {
+        p = asg_arc_pushp(h->ug->g);
+        *p = ug_2->g->arc[i];
+        p->v += (off<<1);
+        p->ul += (off<<33);
+    }
+    free(h->ug->g->idx);
+    h->ug->g->idx = 0;
+    h->ug->g->is_srt = 0;
+    asg_cleanup(h->ug->g);
+    
+
+
+    ///update unitigs
+    ma_utg_t *pu = NULL;
+    for (i = 0; i < ug_2->u.n; i++)
+    {
+        kv_pushp(ma_utg_t, h->ug->u, &pu);
+        *pu = ug_2->u.a[i];
+        ug_2->u.a[i].a = NULL;
+        ug_2->u.a[i].s = NULL;
+    }
+
+    for (i = 0; i < h->ug->g->n_seq; i++)
+    {
+        h->ug->g->seq[i].c = (i < off? FATHER:MOTHER);
+    }
+    /*******************************for debug************************************/
+    if(h->ug->g->n_seq != h->ug->u.n)
+    {
+        fprintf(stderr, "ERROR-non-equal-length\n");
+    }
+    fprintf(stderr, "h->ug->u.n-%u, off-%lu, ug_2->u.n-%u\n", (uint32_t)h->ug->u.n, off, (uint32_t)ug_2->u.n);
+    for (i = 0; i < h->ug->g->n_seq; i++)
+    {
+        if(h->ug->g->seq[i].len != h->ug->u.a[i].len)
+        {
+            fprintf(stderr, "****ERROR-1-%lu: g->seq[i].len-%u, u.a[i].len-%u\n", 
+            i, (uint32_t)h->ug->g->seq[i].len, (uint32_t)h->ug->u.a[i].len);
+        }
+        pu = i < off? (&ug_1->u.a[i]) : (&ug_2->u.a[i - off]);
+        if(h->ug->g->seq[i].len != pu->len)
+        {
+            fprintf(stderr, "ERROR-2\n");
+        } 
+    }
+    
+    /*******************************for debug************************************/
+    ug_1 = NULL;
+    ma_ug_destroy(ug_2);
+}
+
+horder_t *init_horder_t(kvec_pe_hit *i_hits, uint64_t i_hits_uid_bits, uint64_t i_hits_pos_mode, 
+asg_t *i_rg, ma_ug_t* i_ug, bubble_type* bub, ug_opt_t *opt)
+{
+    horder_t *h = NULL; CALLOC(h, 1);
+    get_r_hits(i_hits, &(h->r_hits), i_rg, i_ug, bub, i_hits_uid_bits, i_hits_pos_mode);
+    h->r_g = copy_read_graph(i_rg);
+    generate_haplotypes(h, opt);
+    update_u_hits(&(h->u_hits), &(h->r_hits), h->ug, h->r_g);    
+
+    return h;
+}
+
+void destory_horder_t(horder_t **h)
+{
+    kv_destroy((*h)->r_hits.a);
+    kv_destroy((*h)->r_hits.idx);
+    kv_destroy((*h)->r_hits.occ);
+
+    kv_destroy((*h)->u_hits.a);
+    kv_destroy((*h)->u_hits.idx);
+    kv_destroy((*h)->u_hits.occ);
+
+    ma_ug_destroy((*h)->ug);
+    asg_destroy((*h)->r_g);
+    free((*h));
+}

horder.h

@@ -0,0 +1,14 @@
+#ifndef __HORDER__
+#define __HORDER__
+#include <stdint.h>
+#include "hic.h"
+typedef struct {
+    kvec_pe_hit r_hits, u_hits;
+    ma_ug_t* ug;
+    asg_t* r_g;
+}horder_t;
+
+horder_t *init_horder_t(kvec_pe_hit *i_hits, uint64_t i_hits_uid_bits, uint64_t i_hits_pos_mode, 
+asg_t *i_rg, ma_ug_t* i_ug, bubble_type* bub, ug_opt_t *opt);
+void destory_horder_t(horder_t **h);
+#endif