From 71e91f3fc2b6cbd6f69adf656c0d9f3446bb8eb6 Mon Sep 17 00:00:00 2001
From: chhylp123 <hcheng@jimmy.harvard.edu>
Date: Wed, 28 Apr 2021 20:27:43 -0400
Subject: [PATCH] assgin disconnected parts

---
 CommandLines.h |  2 +-
 Overlaps.cpp   | 11 ++++++-----
 hic.cpp        |  4 +++-
 3 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/CommandLines.h b/CommandLines.h
index b71d83d..04584f8 100644
--- a/CommandLines.h
+++ b/CommandLines.h
@@ -4,7 +4,7 @@
 #include <pthread.h>
 #include <stdint.h>
 
-#define HA_VERSION "0.15.1-r330"
+#define HA_VERSION "0.15.1-r331"
 
 #define VERBOSE 0
 
diff --git a/Overlaps.cpp b/Overlaps.cpp
index 915c447..192d097 100644
--- a/Overlaps.cpp
+++ b/Overlaps.cpp
@@ -11298,8 +11298,9 @@ void refine_u_trans_t(u_trans_hit_t *q, kv_ca_buf_t* cb)
     
 
 
-
-    if(q->tScur >= q->tEcur) fprintf(stderr, "ERROR5\n");
+    ///might be equal
+    if(q->tScur > q->tEcur) fprintf(stderr, "ERROR5\n");
+    // if(q->tScur >= q->tEcur)
     // {
     //     fprintf(stderr, "\n###q->tScur: %u, s: %u, si: %u, q->tEcur: %u, e: %u, ei: %u\n", 
     //     q->tScur, s, si, q->tEcur, e, ei);
@@ -11780,6 +11781,7 @@ ma_ug_t *ug, uint32_t flag, double score, const char* cmd)
         t_ch->c_buf.a[0].c_y_p = t_ch->c_buf.a[1].c_y_p - len;
     }
 
+    ///chain is [s, e)
     if(t_ch->c_buf.a[0].c_x_p != 0 && t_ch->c_buf.a[0].c_y_p != 0) fprintf(stderr, "ERROR1\n");
     if(t_ch->c_buf.a[t_ch->c_buf.n-1].c_x_p!= pri_len && 
                                             t_ch->c_buf.a[t_ch->c_buf.n-1].c_y_p!= aux_len)
@@ -11817,6 +11819,8 @@ ma_ug_t *ug, uint32_t flag, double score, const char* cmd)
     for (i = bn; i < t_ch->k_t_b.n; i++)
     {
         kh = &(t_ch->k_t_b.a[i]);
+        if(kh->qEpre <= kh->qSpre) continue;
+        if(kh->tEpre <= kh->tSpre) continue;
         kv_pushp(u_trans_t, t_ch->k_trans, &kt);
         kt->f = flag; kt->rev = ((kh->qn ^ kh->tn) & 1); kt->del = 0; 
         kt->qn = kh->qn>>1; kt->qs = kh->qSpre; kt->qe = kh->qEpre;
@@ -11831,9 +11835,6 @@ ma_ug_t *ug, uint32_t flag, double score, const char* cmd)
             y_score = ((double)(kt->te-kt->ts)/(double)(t_ch->c_buf.a[t_ch->c_buf.n-1].c_y_p-t_ch->c_buf.a[0].c_y_p))*score;
             kt->nw = MIN(x_score, y_score);
         }
-        
-        // fprintf(stderr, "s-utg%.6ul\t%u\t%u\td-utg%.6ul\t%u\t%u\trev(%u)\n", 
-        // kt->qn+1, kt->qs, kt->qe, kt->tn+1, kt->ts, kt->te, kt->rev);
     }
 
     
diff --git a/hic.cpp b/hic.cpp
index 75b0b99..1cc981f 100644
--- a/hic.cpp
+++ b/hic.cpp
@@ -14598,7 +14598,7 @@ void print_kv_weight(kv_u_trans_t *ta)
 
 void print_debug_hc_links(ha_ug_index* idx, bubble_type* bub, hc_links* lk, kv_u_trans_t *ta, kvec_pe_hit* hits)
 {
-    uint64_t k, len = 0, shif = 64 - idx->uID_bits, beg, end;
+    uint64_t k, len = 0, occ = 0, shif = 64 - idx->uID_bits, beg, end;
     hc_edge *e = NULL;
     for (k = 0; k < hits->a.n; ++k) 
     {
@@ -14608,9 +14608,11 @@ void print_debug_hc_links(ha_ug_index* idx, bubble_type* bub, hc_links* lk, kv_u
         if(IF_HOM(beg, *bub)) continue;
         if(IF_HOM(end, *bub)) continue;
         len += (hits->a.a[k].len>>32) + ((uint32_t)hits->a.a[k].len);
+        occ++;
     }
 
     fprintf(stderr, "# total Hi-C aligned bases: %lu\n", len);
+    fprintf(stderr, "# total Hi-C aligned pairs: %lu\n", occ);
     for (k = 0; k < ta->n; k++)
     {
         e = get_hc_edge(lk, ta->a[k].qn, ta->a[k].tn, 0);