4 files changed, 376 insertions, 0 deletions
diff --git a/src/load/load_asn.c b/src/load/load_asn.c
new file mode 100644
index 0000000..fc27d84
--- a/dev/null
+++ b/src/load/load_asn.c
@@ -0,0 +1,173 @@
+#define _GNU_SOURCE
+#include "load_asn.h"
+#include <string.h>
+#include <asn.h>
+#include <objgbseq.h>
+#include <objsset.h>
+#include <sqnutils.h>
+void
+print_asn (ObjectIdPtr oid, SeqIdPtr id, ValNodePtr descr, SeqAnnotPtr annot)
+{
+  /*
+   * Print the record identifiers.
+   */
+  printf ("  IDENTIFIERS\n");
+  printf ("  -----------\n");
+  while (oid != NULL)
+    {
+      printf("%i, %s\n", oid->id, oid->str);
+    }
+  while (id != NULL)
+    {
+      //      printf ("ID: %i\n", id->choice);
+      char idval[256];
+      SeqIdPrint (id, idval, PRINTID_FASTA_SHORT);
+      printf ("  %s\n", idval);
+      //     if (id->choice == SEQID_GI)
+              //printf ("GI: %i\n", id->data.intvalue);
+ 
+      id = id->next;
+    }
+  /*
+   * Print descriptions.
+   * [http://www.ncbi.nlm.nih.gov/IEB/ToolBox/SDKDOCS/BIOSEQ.HTML#_Seq-descr:_Describing_the]
+   */
+  printf ("\n  DESCRIPTIONS\n");
+  printf ("  ------------\n");
+  while (descr != NULL)
+    {
+      switch (descr->choice)
+        {
+        case Seq_descr_title:
+          printf ("  TITLE: %s\n", (char*)descr->data.ptrvalue);
+          break;
+        case Seq_descr_genbank:
+          printf ("  GENBANK\n");
+          break;
+        case Seq_descr_pub:
+          printf ("  PUB\n");
+          break;
+        case Seq_descr_create_date:
+          printf ("  CREATE DATE\n");
+          break;
+        case Seq_descr_update_date:
+          printf ("  UPDATE DATE\n");
+          break;
+        case Seq_descr_source:
+          printf ("  BIOSOURCE\n");
+          break;
+        case Seq_descr_molinfo:
+          printf ("  MOLINFO\n");
+          break;
+        default:
+          printf ("  DESCRIPTION CHOICE=%i\n", descr->choice);
+          break;
+        }
+      descr = descr->next;
+    }
+  /*
+   * Print annotations.
+   */
+  printf ("\n  ANNOTATIONS\n");
+  printf ("  -----------\n");
+  while (annot != NULL)
+    {
+      printf ("  ANNOTATION: %s, ", annot->name);
+      if (annot->desc != NULL) {
+        switch (annot->desc->choice)
+          {
+          case Annot_descr_name:
+            printf ("  NAME: %s\n", (char*)annot->desc->data.ptrvalue);
+            break;
+          default:
+            printf ("  CHOICE=%i\n", annot->desc->choice);
+            break;
+          }
+      }
+      else
+        printf ("  NONE\n");
+      annot = annot->next;
+    }
+}
+/*
+ * Based on example at
+ * [http://www.ncbi.nlm.nih.gov/IEB/ToolBox/SDKDOCS/SEQUTIL.HTML].
+ */
+void
+load_asn (hid_t file_id, const char* file_name)
+{
+  char* asn_file = strdup(file_name);
+  AsnIoPtr aip = AsnIoOpen (asn_file, "r"); 
+  SeqEntryPtr sep = SeqEntryAsnRead (aip, NULL);
+  BioseqSetPtr bsetp = 0;
+  ValNodePtr descr = 0;
+  SeqAnnotPtr annot = 0;
+  SeqIdPtr id = 0;
+  ObjectIdPtr oid = 0;
+  /*
+   * Data file statistics.
+   */
+  printf ("NODES: %i\tBIOSEQS: %i\n", ValNodeLen (sep), BioseqCount (sep));
+  printf ("\n");
+  /*
+   * This loop needs to be corrected to handle nesting of sets.
+   */
+  while (sep != NULL)
+    {
+      bsetp = (BioseqSetPtr) sep->data.ptrvalue;  
+      if (bsetp != NULL)
+        {
+          oid = bsetp->id;
+          id = NULL;
+          descr = bsetp->descr;
+          annot = bsetp->annot;
+        }
+      printf ("BIOSEQSET\n");
+      printf ("\n");
+      print_asn (oid, id, descr, annot);
+      printf ("\n");
+      
+      /*
+       * Process Bioseqs in the set.
+       */
+      SeqEntryPtr sep2 = bsetp->seq_set;
+      while (sep2 != NULL)
+        {
+          BioseqPtr bsp = sep2->data.ptrvalue;
+          if (bsp != NULL)
+            {
+              oid = NULL;
+              id = bsp->id;
+              descr = bsp->descr;
+              annot = bsp->annot;
+              
+              printf ("BIOSEQ\n");
+                      printf ("\n");      
+              print_asn (oid, id, descr, annot);
+              printf ("\n");
+            }
+          sep2 = sep2->next;
+        }
+      
+      sep = sep->next;
+    } 
+  AsnIoClose (aip);
+  free (asn_file);
+  return;
+}

diff --git a/src/load/load_asn.c b/src/load/load_asn.c new file mode 100644 index 0000000..fc27d84 --- a/dev/null +++ b/src/load/load_asn.c
@@ -0,0 +1,173 @@
	1	#define _GNU_SOURCE
	2	#include "load_asn.h"
	3	#include <string.h>
	4	#include <asn.h>
	5	#include <objgbseq.h>
	6	#include <objsset.h>
	7	#include <sqnutils.h>
	8
	9	void
	10	print_asn (ObjectIdPtr oid, SeqIdPtr id, ValNodePtr descr, SeqAnnotPtr annot)
	11	{
	12	/*
	13	* Print the record identifiers.
	14	*/
	15	printf (" IDENTIFIERS\n");
	16	printf (" -----------\n");
	17	while (oid != NULL)
	18	{
	19	printf("%i, %s\n", oid->id, oid->str);
	20	}
	21	while (id != NULL)
	22	{
	23	// printf ("ID: %i\n", id->choice);
	24
	25	char idval[256];
	26	SeqIdPrint (id, idval, PRINTID_FASTA_SHORT);
	27	printf (" %s\n", idval);
	28
	29	// if (id->choice == SEQID_GI)
	30	//printf ("GI: %i\n", id->data.intvalue);
	31
	32	id = id->next;
	33	}
	34
	35	/*
	36	* Print descriptions.
	37	* [http://www.ncbi.nlm.nih.gov/IEB/ToolBox/SDKDOCS/BIOSEQ.HTML#_Seq-descr:_Describing_the]
	38	*/
	39	printf ("\n DESCRIPTIONS\n");
	40	printf (" ------------\n");
	41	while (descr != NULL)
	42	{
	43	switch (descr->choice)
	44	{
	45	case Seq_descr_title:
	46	printf (" TITLE: %s\n", (char*)descr->data.ptrvalue);
	47	break;
	48	case Seq_descr_genbank:
	49	printf (" GENBANK\n");
	50	break;
	51	case Seq_descr_pub:
	52	printf (" PUB\n");
	53	break;
	54	case Seq_descr_create_date:
	55	printf (" CREATE DATE\n");
	56	break;
	57	case Seq_descr_update_date:
	58	printf (" UPDATE DATE\n");
	59	break;
	60	case Seq_descr_source:
	61	printf (" BIOSOURCE\n");
	62	break;
	63	case Seq_descr_molinfo:
	64	printf (" MOLINFO\n");
	65	break;
	66	default:
	67	printf (" DESCRIPTION CHOICE=%i\n", descr->choice);
	68	break;
	69	}
	70
	71	descr = descr->next;
	72	}
	73
	74	/*
	75	* Print annotations.
	76	*/
	77	printf ("\n ANNOTATIONS\n");
	78	printf (" -----------\n");
	79	while (annot != NULL)
	80	{
	81	printf (" ANNOTATION: %s, ", annot->name);
	82	if (annot->desc != NULL) {
	83	switch (annot->desc->choice)
	84	{
	85	case Annot_descr_name:
	86	printf (" NAME: %s\n", (char*)annot->desc->data.ptrvalue);
	87	break;
	88	default:
	89	printf (" CHOICE=%i\n", annot->desc->choice);
	90	break;
	91	}
	92	}
	93	else
	94	printf (" NONE\n");
	95
	96	annot = annot->next;
	97	}
	98
	99	}
	100
	101	/*
	102	* Based on example at
	103	* [http://www.ncbi.nlm.nih.gov/IEB/ToolBox/SDKDOCS/SEQUTIL.HTML].
	104	*/
	105	void
	106	load_asn (hid_t file_id, const char* file_name)
	107	{
	108	char* asn_file = strdup(file_name);
	109	AsnIoPtr aip = AsnIoOpen (asn_file, "r");
	110	SeqEntryPtr sep = SeqEntryAsnRead (aip, NULL);
	111	BioseqSetPtr bsetp = 0;
	112	ValNodePtr descr = 0;
	113	SeqAnnotPtr annot = 0;
	114	SeqIdPtr id = 0;
	115	ObjectIdPtr oid = 0;
	116
	117	/*
	118	* Data file statistics.
	119	*/
	120	printf ("NODES: %i\tBIOSEQS: %i\n", ValNodeLen (sep), BioseqCount (sep));
	121	printf ("\n");
	122
	123	/*
	124	* This loop needs to be corrected to handle nesting of sets.
	125	*/
	126
	127	while (sep != NULL)
	128	{
	129	bsetp = (BioseqSetPtr) sep->data.ptrvalue;
	130	if (bsetp != NULL)
	131	{
	132	oid = bsetp->id;
	133	id = NULL;
	134	descr = bsetp->descr;
	135	annot = bsetp->annot;
	136	}
	137
	138	printf ("BIOSEQSET\n");
	139	printf ("\n");
	140	print_asn (oid, id, descr, annot);
	141	printf ("\n");
	142
	143	/*
	144	* Process Bioseqs in the set.
	145	*/
	146	SeqEntryPtr sep2 = bsetp->seq_set;
	147	while (sep2 != NULL)
	148	{
	149	BioseqPtr bsp = sep2->data.ptrvalue;
	150	if (bsp != NULL)
	151	{
	152	oid = NULL;
	153	id = bsp->id;
	154	descr = bsp->descr;
	155	annot = bsp->annot;
	156
	157	printf ("BIOSEQ\n");
	158	printf ("\n");
	159	print_asn (oid, id, descr, annot);
	160	printf ("\n");
	161	}
	162
	163	sep2 = sep2->next;
	164	}
	165
	166	sep = sep->next;
	167	}
	168
	169	AsnIoClose (aip);
	170	free (asn_file);
	171
	172	return;
	173	}