save hit_start hit_stop

EBI-Metagenomics · Feb 15, 2024 · 25919b7 · 25919b7
1 parent 78bafe9
commit 25919b7
Show file tree

Hide file tree

Showing 9 changed files with 25 additions and 10 deletions.
diff --git a/c-core/CMakeLists.txt b/c-core/CMakeLists.txt
@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.20.2 FATAL_ERROR)
-project(deciphon VERSION 0.17.2 LANGUAGES C)
+project(deciphon VERSION 0.17.3 LANGUAGES C)
 
 include(cmake/warnings.cmake)
 include(cmake/sanitizers.cmake)

diff --git a/c-core/product.c b/c-core/product.c
@@ -64,6 +64,8 @@ int product_close(struct product *x)
   ok &= fputs("window\t", fp) >= 0;
   ok &= fputs("window_start\t", fp) >= 0;
   ok &= fputs("window_stop\t", fp) >= 0;
+  ok &= fputs("hit_start\t", fp) >= 0;
+  ok &= fputs("hit_stop\t", fp) >= 0;
   ok &= fputs("profile\t", fp) >= 0;
   ok &= fputs("abc\t", fp) >= 0;
   ok &= fputs("lrt\t", fp) >= 0;

diff --git a/c-core/product_line.c b/c-core/product_line.c
@@ -13,6 +13,9 @@ void product_line_init(struct product_line *x)
   x->window_start = 0;
   x->window_stop = 0;
 
+  x->hit_start = 0;
+  x->hit_stop = 0;
+
   memset(x->protein, 0, sizeof_field(struct product_line, protein));
   memset(x->abc, 0, sizeof_field(struct product_line, abc));
 

diff --git a/c-core/product_line.h b/c-core/product_line.h
@@ -6,9 +6,14 @@ struct product_line
   long sequence;
 
   int window;
+  // [window_start, window_stop)
   int window_start;
   int window_stop;
 
+  // [hit_start, hit_stop)
+  int hit_start;
+  int hit_stop;
+
   char protein[64];
   char abc[16];
 

diff --git a/c-core/product_thread.c b/c-core/product_thread.c
@@ -38,6 +38,8 @@ int product_thread_put_match(struct product_thread *x, struct match *match,
   if (fprintf(fp, "%d\t", line->window) < 0) defer_error(DCP_EWRITEPROD);
   if (fprintf(fp, "%d\t", line->window_start) < 0) defer_error(DCP_EWRITEPROD);
   if (fprintf(fp, "%d\t", line->window_stop) < 0) defer_error(DCP_EWRITEPROD);
+  if (fprintf(fp, "%d\t", line->hit_start) < 0) defer_error(DCP_EWRITEPROD);
+  if (fprintf(fp, "%d\t", line->hit_stop) < 0) defer_error(DCP_EWRITEPROD);
   if (fprintf(fp, "%s\t", line->protein) < 0) defer_error(DCP_EWRITEPROD);
   if (fprintf(fp, "%s\t", line->abc) < 0) defer_error(DCP_EWRITEPROD);
   if (fprintf(fp, "%.1f\t", line->lrt) < 0) defer_error(DCP_EWRITEPROD);

diff --git a/c-core/test_massive.c b/c-core/test_massive.c
@@ -31,7 +31,7 @@ int main(void)
   }
   eq(scan_run(scan, PRODDIR, NULL, NULL), 0);
   eq(scan_progress(scan), 100);
-  eq(chksum(PRODDIR "/products.tsv"), 27703);
+  eq(chksum(PRODDIR "/products.tsv"), 61836);
   eq(scan_close(scan), 0);
 
   scan_del(scan);

diff --git a/c-core/test_scan.c b/c-core/test_scan.c
@@ -15,8 +15,8 @@ static struct params params_list[] = {
     {1, false, false}, {1, false, false}, {1, false, true}, {1, false, true},
     {1, true, false},  {1, true, false},  {1, true, true},  {1, true, true}};
 static bool dial_list[] = {true, false, true, false, true, false, true, false};
-static long chksum_list[] = {55932, 55932, 35953, 35953,
-                             4940,  4940,  4940,  4940};
+static long chksum_list[] = {10096, 10096, 27848, 27848,
+                             44180, 44180, 44180, 44180};
 
 static void test_invalid_sequence();
 static void test_normal_scan(void);

diff --git a/c-core/test_window.c b/c-core/test_window.c
@@ -40,7 +40,7 @@ int main(void)
   eq(scan_add(scan, sequences[0].id, sequences[0].name, seq), 0);
   eq(scan_run(scan, PRODDIR, NULL, NULL), 0);
   eq(scan_progress(scan), 100);
-  eq(chksum(PRODDIR "/products.tsv"), 25112);
+  eq(chksum(PRODDIR "/products.tsv"), 38063);
   eq(scan_close(scan), 0);
 
   scan_del(scan);

diff --git a/c-core/thread.c b/c-core/thread.c
@@ -134,7 +134,7 @@ static int code_fn(int pos, int len, void *arg)
 }
 
 static int trim_path(struct protein *, struct imm_seq const *,
-                     struct imm_path *, struct imm_seq *);
+                     struct imm_path *, struct imm_seq *, int *seqstart);
 
 static int process_window(struct thread *x, int protein_idx,
                           struct window const *w)
@@ -174,8 +174,11 @@ static int process_window(struct thread *x, int protein_idx,
     subseq = seq->imm.seq;
   else
   {
-    if ((rc = trim_path(&x->protein, &seq->imm.seq, subpath, &subseq)))
+    line->hit_start = 0;
+    if ((rc = trim_path(&x->protein, &seq->imm.seq, subpath, &subseq,
+                        &line->hit_start)))
       return rc;
+    line->hit_stop = line->hit_start + imm_seq_size(&subseq);
   }
 
   if (hmmer_online(&x->hmmer))
@@ -206,7 +209,7 @@ static int process_window(struct thread *x, int protein_idx,
 }
 
 static int trim_path(struct protein *protein, struct imm_seq const *seq,
-                     struct imm_path *path, struct imm_seq *subseq)
+                     struct imm_path *path, struct imm_seq *subseq, int *seqstart)
 {
   int rc = 0;
 
@@ -225,7 +228,7 @@ static int trim_path(struct protein *protein, struct imm_seq const *seq,
   }
   if (rc) return rc;
   start = match_iter_tell(&it) - 1;
-  int seqstart = match_iter_seqtell(&it) - match.step.seqsize;
+  *seqstart = match_iter_seqtell(&it) - match.step.seqsize;
 
   if ((rc = match_iter_seek(&it, &match, INT_MAX))) return rc;
   while (!(rc = match_iter_prev(&it, &match)))
@@ -239,7 +242,7 @@ static int trim_path(struct protein *protein, struct imm_seq const *seq,
 
   match_iter_seek(&it, &match, start);
 
-  *subseq = imm_seq_slice(seq, imm_range(seqstart, seqstop));
+  *subseq = imm_seq_slice(seq, imm_range(*seqstart, seqstop));
   imm_path_cut(path, start, stop - start);
 
   return 0;