bed_sort created

git-svn-id: http://biopieces.googlecode.com/svn/trunk@256 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/code_c/Maasha/src/Makefile b/code_c/Maasha/src/Makefile
index 527b6eab9ba72dc896843d2c1116aa25f4987fc9..4312b1fd7bfc2b5709207781704768e928ac42c8 100644 (file)
--- a/code_c/Maasha/src/Makefile
+++ b/code_c/Maasha/src/Makefile
@@ -11,7 +11,7 @@ TEST_DIR = test/
 INC = -I $(INC_DIR)
 LIB = -lm $(LIB_DIR)*.o
 
-all: libs utest bipartite_scan fasta_count repeat-O-matic
+all: libs utest bed_sort bipartite_scan fasta_count repeat-O-matic
 
 libs:
        cd $(LIB_DIR) && ${MAKE} all
@@ -19,6 +19,9 @@ libs:
 utest:
        cd $(TEST_DIR) && ${MAKE} all
 
+bed_sort: bed_sort.c
+       $(CC) $(Cflags) $(INC) $(LIB) bed_sort.c -o bed_sort
+
 bipartite_scan: bipartite_scan.c
        $(CC) $(Cflags) $(INC) $(LIB) bipartite_scan.c -o bipartite_scan
 
@@ -31,6 +34,7 @@ repeat-O-matic: repeat-O-matic.c
 clean:
        cd $(LIB_DIR) && ${MAKE} clean
        cd $(TEST_DIR) && ${MAKE} clean
+       rm bed_sort
        rm bipartite_scan
        rm fasta_count
        rm repeat-O-matic

diff --git a/code_c/Maasha/src/bed_sort.c b/code_c/Maasha/src/bed_sort.c
new file mode 100644 (file)
index 0000000..d8e84eb
--- /dev/null
+++ b/code_c/Maasha/src/bed_sort.c
@@ -0,0 +1,22 @@
+/* Martin Asser Hansen (mail@maasha.dk) Copyright (C) 2008 - All right reserved */
+
+#include "common.h"
+#include "list.h"
+#include "ucsc.h"
+
+
+int main( int argc, char *argv[] )
+{
+    char    *file    = NULL;
+    list_sl *entries = NULL;
+
+    file = argv[ 1 ];
+
+    entries = bed_entries_get( file, 3 );
+
+    list_sl_sort( &entries, cmp_bed_sort_chr_beg );
+
+    bed_entries_put( entries, 3 );
+
+    return EXIT_SUCCESS;
+}

diff --git a/code_c/Maasha/src/inc/ucsc.h b/code_c/Maasha/src/inc/ucsc.h
index 0ca029d142e9493788dfa6d71d5f669e9d8c4d89..6e655c58d3fc3cdd35b9c5c4400232dc6349de5b 100644 (file)
--- a/code_c/Maasha/src/inc/ucsc.h
+++ b/code_c/Maasha/src/inc/ucsc.h
@@ -1,11 +1,11 @@
 /* Martin Asser Hansen (mail@maasha.dk) Copyright (C) 2008 - All right reserved */
 
 #define BED_BUFFER         2048
-#define BED_CHR_MAX          64
+#define BED_CHR_MAX          16
 #define BED_QID_MAX         256
 #define BED_ITEMRGB_MAX      16
-#define BED_BLOCKSIZES_MAX  512
-#define BED_QBEGS_MAX       512
+#define BED_BLOCKSIZES_MAX  256
+#define BED_QBEGS_MAX       256
 
 struct _bed_entry
 {
@@ -14,7 +14,7 @@ struct _bed_entry
     uint   chr_beg;
     uint   chr_end;
     char  *q_id;
-    uint   score;
+    int    score;
     char   strand; 
     uint   thick_beg;
     uint   thick_end;
@@ -31,4 +31,6 @@ bed_entry *bed_entry_get( FILE *fp, const int cols );
 list_sl   *bed_entries_get( char *path, const int cols );
 void       bed_entry_put( bed_entry *entry, int cols );
 void       bed_entries_put( list_sl *entries, int cols );
-int        cmp_bed3_entries_sort( const void *a, const void *b );
+int        cmp_bed_sort_beg( const void *a, const void *b );
+int        cmp_bed_sort_chr_beg( const void *a, const void *b );
+int        cmp_bed_sort_chr_strand_beg( const void *a, const void *b );

diff --git a/code_c/Maasha/src/lib/ucsc.c b/code_c/Maasha/src/lib/ucsc.c
index 6fab258dbb79e881eac2f49328340e31e0066e2d..bdb722a85b3ce98d713adb85d8ad97ca0e1ec9b2 100644 (file)
--- a/code_c/Maasha/src/lib/ucsc.c
+++ b/code_c/Maasha/src/lib/ucsc.c
@@ -97,7 +97,7 @@ bed_entry *bed_entry_get( FILE *fp, int cols )
         {
             sscanf(
                     buffer,
-                    "%s\t%u\t%u\t%s\t%u",
+                    "%s\t%u\t%u\t%s\t%i",
                     entry->chr, 
                     &entry->chr_beg, 
                     &entry->chr_end,
@@ -112,7 +112,7 @@ bed_entry *bed_entry_get( FILE *fp, int cols )
         {
             sscanf(
                     buffer,
-                    "%s\t%u\t%u\t%s\t%u\t%c",
+                    "%s\t%u\t%u\t%s\t%i\t%c",
                     entry->chr, 
                     &entry->chr_beg, 
                     &entry->chr_end,
@@ -128,7 +128,7 @@ bed_entry *bed_entry_get( FILE *fp, int cols )
         {
             sscanf(
                     buffer,
-                    "%s\t%u\t%u\t%s\t%u\t%c\t%u\t%u\t%s\t%u\t%s\t%s",
+                    "%s\t%u\t%u\t%s\t%i\t%c\t%u\t%u\t%s\t%u\t%s\t%s",
                     entry->chr, 
                     &entry->chr_beg, 
                     &entry->chr_end,
@@ -215,7 +215,7 @@ void bed_entry_put( bed_entry *entry, int cols )
     else if ( cols == 5 )
     {
         printf(
-            "%s\t%u\t%u\t%s\t%u\n",
+            "%s\t%u\t%u\t%s\t%i\n",
             entry->chr,
             entry->chr_beg,
             entry->chr_end,
@@ -226,7 +226,7 @@ void bed_entry_put( bed_entry *entry, int cols )
     else if ( cols == 6 )
     {
         printf(
-            "%s\t%u\t%u\t%s\t%u\t%c\n",
+            "%s\t%u\t%u\t%s\t%i\t%c\n",
             entry->chr,
             entry->chr_beg,
             entry->chr_end,
@@ -238,7 +238,7 @@ void bed_entry_put( bed_entry *entry, int cols )
     else if ( cols == 12 )
     {
         printf(
-            "%s\t%u\t%u\t%s\t%u\t%c\t%u\t%u\t%s\t%u\t%s\t%s\n",
+            "%s\t%u\t%u\t%s\t%i\t%c\t%u\t%u\t%s\t%u\t%s\t%s\n",
             entry->chr,
             entry->chr_beg,
             entry->chr_end,
@@ -272,7 +272,7 @@ void bed_entries_put( list_sl *entries, int cols )
 }
 
 
-int cmp_bed3_entries_sort( const void *a, const void *b )
+int cmp_bed_sort_beg( const void *a, const void *b )
 {
     node_sl *a_node = *( ( node_sl ** ) a );
     node_sl *b_node = *( ( node_sl ** ) b );
@@ -280,10 +280,66 @@ int cmp_bed3_entries_sort( const void *a, const void *b )
     bed_entry *a_entry = ( bed_entry * ) a_node->val;
     bed_entry *b_entry = ( bed_entry * ) b_node->val;
 
-    if ( a_entry->chr_end < b_entry->chr_end ) {
+    if ( a_entry->chr_beg < b_entry->chr_beg ) {
+        return -1;
+    } else if ( a_entry->chr_beg > b_entry->chr_beg ) {
+        return 1;
+    } else {
+        return 0;
+    }
+}
+
+
+int cmp_bed_sort_chr_beg( const void *a, const void *b )
+{
+    node_sl *a_node = *( ( node_sl ** ) a );
+    node_sl *b_node = *( ( node_sl ** ) b );
+
+    bed_entry *a_entry = ( bed_entry * ) a_node->val;
+    bed_entry *b_entry = ( bed_entry * ) b_node->val;
+
+    int diff = 0;
+
+    diff = strcmp( a_entry->chr, b_entry->chr );
+
+    if ( diff < 0 ) {
+        return -1;
+    } else if ( diff > 0 ) {
+        return 1;
+    } else if ( a_entry->chr_beg < b_entry->chr_beg ) {
+        return -1;
+    } else if ( a_entry->chr_beg > b_entry->chr_beg ) {
         return 1;
-    } else if ( a_entry->chr_end > b_entry->chr_end ) {
+    } else {
+        return 0;
+    }
+}
+
+
+int cmp_bed_sort_chr_strand_beg( const void *a, const void *b )
+{
+    node_sl *a_node = *( ( node_sl ** ) a );
+    node_sl *b_node = *( ( node_sl ** ) b );
+
+    bed_entry *a_entry = ( bed_entry * ) a_node->val;
+    bed_entry *b_entry = ( bed_entry * ) b_node->val;
+
+    int diff = 0;
+
+    diff = strcmp( a_entry->chr, b_entry->chr );
+
+    if ( diff < 0 ) {
         return -1;
+    } else if ( diff > 0 ) {
+        return 1;
+    } else if ( a_entry->strand < b_entry->strand ) {
+        return -1;
+    } else if ( a_entry->strand > b_entry->strand ) {
+        return 1;
+    } else if ( a_entry->chr_beg < b_entry->chr_beg ) {
+        return -1;
+    } else if ( a_entry->chr_beg > b_entry->chr_beg ) {
+        return 1;
     } else {
         return 0;
     }

diff --git a/code_c/Maasha/src/test/test_files/test12.bed b/code_c/Maasha/src/test/test_files/test12.bed
index 7e60b265f7a5abe8938553771293dad78fb55481..79f75f79cc8cbfd819ec790f3d2b0d6767c8a74e 100644 (file)
--- a/code_c/Maasha/src/test/test_files/test12.bed
+++ b/code_c/Maasha/src/test/test_files/test12.bed
@@ -1,10 +1,10 @@
-chr4   31176   31602   AA695812        0       -       31176   31602   0       1       426,    0,
-chr4   44448   44874   AA695812        0       -       44448   44874   0       1       426,    0,
-chr4   50522   50841   AA142091        0       -       50522   50841   0       2       81,237, 0,82,
-chr4   57489   57808   AA142091        0       -       57489   57808   0       2       81,237, 0,82,
-chr4   59352   59778   AA695812        0       -       59352   59778   0       1       426,    0,
-chr4   63580   64332   AA979544        0       -       63580   64332   0       1       752,    0,
-chr4   63710   64332   AA979534        0       -       63710   64332   0       3       111,481,30,     0,111,592,
-chr4   70946   71196   AA699063        0       -       70946   71196   0       2       142,55, 0,195,
+chr14  31176   31602   AA695812        0       -       31176   31602   0       1       426,    0,
+chr4   70946   71196   AA699063        0       +       70946   71196   0       2       142,55, 0,195,
 chr4   72831   76893   AA264101        0       -       72831   76893   0       2       179,437,        0,3625,
 chr4   72872   76630   AA694817        0       -       72872   76630   0       3       83,54,174,      0,84,3584,
+chr4   50522   50841   AA142091        0       -       50522   50841   0       2       81,237, 0,82,
+chr14  44448   44874   AA695812        0       -       44448   44874   0       1       426,    0,
+chr14  57489   57808   AA142091        0       +       57489   57808   0       2       81,237, 0,82,
+chr4   59352   59778   AA695812        0       -       59352   59778   0       1       426,    0,
+chr4   63580   64332   AA979544        0       +       63580   64332   0       1       752,    0,
+chr24  63710   64332   AA979534        0       -       63710   64332   0       3       111,481,30,     0,111,592,

diff --git a/code_c/Maasha/src/test/test_ucsc.c b/code_c/Maasha/src/test/test_ucsc.c
index 2ef6220bd8982e8abe743e173d052faed3783994..4ef1cd5f74c4d878debba5d45d5a9e8529a2f83e 100644 (file)
--- a/code_c/Maasha/src/test/test_ucsc.c
+++ b/code_c/Maasha/src/test/test_ucsc.c
@@ -67,7 +67,8 @@ void test_bed_entries_sort()
     
     entries = bed_entries_get( path, 0 );
 
-    list_sl_sort( &entries, cmp_bed3_entries_sort );
+    list_sl_sort( &entries, cmp_bed_sort_chr_beg );
+    list_sl_sort( &entries, cmp_bed_sort_chr_strand_beg );
 
     bed_entries_put( entries, 0 );