Beginning of implmentation to iterate through and existing HDF5 and add calculate protein type field values based on BLAST queries. This code currently does not compile.

3 files changed, 59 insertions, 14 deletions

diff --git a/src/assign_protein_type.c b/src/assign_protein_type.c
index 54db84e..1b58f54 100644
--- a/src/assign_protein_type.c
+++ b/src/assign_protein_type.c
@@ -1,10 +1,12 @@
 #include "assign_protein_type.h"
 #include "check_ncbi_error.h"
+#include "check_h5_error.h"
 #include <ncbi.h>
 #include <readdb.h>
 #include <blast.h>
 #include <salpacc.h>
 #include <stdbool.h>
+#include <hdf5_hl.h>
 
 /*
  * BLAST database containing all of the influenza protein sequences.
@@ -45,6 +47,33 @@ assign_protein_type (hid_t file_id)
   ValNodePtr error_returns = NULL;
 
   /*
+   * Read the data from HDF5 file.
+   */
+  hsize_t nfields;
+  hsize_t nrecords;
+  herr_t status = H5TBget_table_info (file_id, "influenza.faa", &nfields,
+                                      &nrecords);
+  if (status < 0)
+    check_h5_error (status, __FILE__, __LINE__);
+
+  /*
+   * todo: Allocate memory of nrecords for dst_buf.
+   *
+   * todo: Refactor code to share structres in read and write HDF5
+   * calls.
+   */
+
+  status = H5TBread_table (file_id, "influenza.faa", dst_size, dst_offset,
+                           dst_sizes, dst_buf);
+  if (status < 0)
+    check_h5_error (status, __FILE__, __LINE__);
+
+  for (int i = 0; i < nrecords; i++)
+    {
+      
+    }
+
+  /*
    * Read the sequence from the database by GI.
    */
   Int4 sequence_number = readdb_gi2seq (seqdb, 453644, NULL);
diff --git a/src/load_influenza_aa_dat.c b/src/load_influenza_aa_dat.c
index f0d9ee5..aed33e8 100644
--- a/src/load_influenza_aa_dat.c
+++ b/src/load_influenza_aa_dat.c
@@ -8,7 +8,7 @@
 #include "load_influenza_aa_dat.h"
 #include "check_error.h"
 #include "check_h5_error.h"
-#include "hdf5_hl.h"
+#include <hdf5_hl.h>
 #include <string.h>
 #include <stdlib.h>
 
diff --git a/src/load_influenza_faa.c b/src/load_influenza_faa.c
index 61bb99d..749b7ad 100644
--- a/src/load_influenza_faa.c
+++ b/src/load_influenza_faa.c
@@ -5,6 +5,8 @@
 #include <string.h>
 #include <stdlib.h>
 
+#define SEQUENCE_DATA_FIELD_NUM 4
+
 void
 load_influenza_faa (hid_t file_id)
 {
@@ -13,24 +15,27 @@ load_influenza_faa (hid_t file_id)
     int gi;
     char gb[9];
     char description[196];
+    char protein_type[7];
   } sequence_data;
-
+  
   size_t dst_size = sizeof (sequence_data);
-  size_t dst_offset[3] =
+  size_t dst_offset[SEQUENCE_DATA_FIELD_NUM] =
     { HOFFSET (sequence_data, gi),
       HOFFSET (sequence_data, gb),
-      HOFFSET (sequence_data, description)
+      HOFFSET (sequence_data, description),
+      HOFFSET (sequence_data, protein_type)
     };
 
   sequence_data dst_buf[1];
 
-  size_t dst_sizes[3] = {
+  size_t dst_sizes[SEQUENCE_DATA_FIELD_NUM] = {
     sizeof (dst_buf[0].gi),
     sizeof (dst_buf[0].gb),
-    sizeof (dst_buf[0].description)
+    sizeof (dst_buf[0].description),
+    sizeof (dst_buf[0].protein_type)
   };
 
-  hid_t field_type[3];
+  hid_t field_type[SEQUENCE_DATA_FIELD_NUM];
 
   field_type[0] = H5T_NATIVE_INT;
 
@@ -42,9 +47,15 @@ load_influenza_faa (hid_t file_id)
   H5Tset_size (description_type, 196);
   field_type[2] = description_type;
 
-  const char *field_names[3] = { "GI",
-                                 "GB",
-                                 "Description" };
+  hid_t protein_type_type = H5Tcopy (H5T_C_S1);
+  H5Tset_size (protein_type_type, 7);
+  field_type[3] = protein_type_type;
+
+  const char *field_names[SEQUENCE_DATA_FIELD_NUM] = 
+    { "GI",
+      "GB",
+      "Description",
+      "Protein Type" };
 
   hsize_t chunk_size = 10;
   int *fill_data = NULL;
@@ -86,13 +97,17 @@ load_influenza_faa (hid_t file_id)
           strncpy (p_data.description, strsep (&running, "|"),
                    sizeof (p_data.description));
 
+          strncpy (p_data.protein_type, "", sizeof (p_data.protein_type));
+
           if (current_line == 1)
             {
               herr_t status = H5TBmake_table ("influenza.faa", file_id,
-                                              "influenza.faa", 3, 1, dst_size,
-                                              field_names, dst_offset,
-                                              field_type, chunk_size,
-                                              fill_data, compress, &p_data);
+                                              "influenza.faa",
+                                              SEQUENCE_DATA_FIELD_NUM, 1,
+                                              dst_size, field_names,
+                                              dst_offset, field_type,
+                                              chunk_size, fill_data, compress,
+                                              &p_data);
               if (status < 0)
                 check_h5_error (status, __FILE__, __LINE__);
             }
@@ -119,6 +134,7 @@ load_influenza_faa (hid_t file_id)
 
   H5Tclose (gb_type);
   H5Tclose (description_type);
+  H5Tclose (protein_type_type);
 
   return;
 }