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;
+}