Modified to take the location of the exp004 directory from an

environment variable.

4 files changed, 60 insertions, 14 deletions

diff --git a/src/aggregator.c b/src/aggregator.c
index c789502..8057bb3 100644
--- a/src/aggregator.c
+++ b/src/aggregator.c
@@ -8,15 +8,25 @@
 #include "load/load_influenza_aa_dat.h"
 #include "load/load_influenza_faa.h"
 #include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
 
 #define BLASTFILE "run20090807.del"
 #define H5FILE "influenza.h5"
-#define INFLUENZA_AA_DAT "/u/ac/dpellegr/Documents/exp004/genomes/INFLUENZA/influenza_aa.dat"
-#define INFLUENZA_FAA "/u/ac/dpellegr/Documents/exp004/genomes/INFLUENZA/influenza.faa"
+#define INFLUENZA_AA_DAT "/genomes/INFLUENZA/influenza_aa.dat"
+#define INFLUENZA_FAA "/genomes/INFLUENZA/influenza.faa"
 
 int
 main ()
 {
+  char* exp004 = getenv ("EXP004");
+  if (exp004 == NULL)
+    {
+      printf ("EXP004 environment variable not set.  This should be set to the "
+              "directory containing the genomes/INFLUENZA data.");
+      exit (0);
+    }
+
   /*
    * Create a new HDF5 file if it does not already exist.  If an
    * existing file is found then open it.
@@ -40,15 +50,26 @@ main ()
   /*
    * Load the supplementary protein data file.
    */
+  char* loc1 = malloc (strlen (exp004) + strlen (INFLUENZA_AA_DAT) + 1);
+  loc1[0] = '\0';
+  strcat (loc1, exp004);
+  strcat (loc1, INFLUENZA_AA_DAT);
   printf ("Loading \"influenza_aa.dat\" with contents of %s.\n",
-          INFLUENZA_AA_DAT);
-  load_influenza_aa_dat (file_id, INFLUENZA_AA_DAT);
+          loc1);
+  load_influenza_aa_dat (file_id, loc1);
+  free (loc1);
 
   /*
    * Load the FASTA protein sequence data file.
    */
-  printf ("Loading \"influenza.faa\" with contents of %s.\n", INFLUENZA_FAA);
-  load_influenza_faa (file_id, INFLUENZA_FAA);
+  char* loc2 = malloc (strlen (exp004) + strlen (INFLUENZA_FAA) + 1);
+  loc2[0] = '\0';
+  strcat (loc2, exp004);
+  strcat (loc2, INFLUENZA_FAA);
+  printf ("Loading \"influenza.faa\" with contents of %s.\n",
+          loc2);
+  load_influenza_faa (file_id, loc2);
+  free (loc2);
 
   /*
    * Load the BLAST scores.
diff --git a/src/assign/assign_blast_scores.c b/src/assign/assign_blast_scores.c
index 7c0493b..3bdcccc 100644
--- a/src/assign/assign_blast_scores.c
+++ b/src/assign/assign_blast_scores.c
@@ -30,7 +30,16 @@ assign_blast_scores (hid_t file_id)
    * Write the contents of the matrix to a HDF5 file.
    */
   PetscViewer viewer;
-  PetscErrorCode err = PetscViewerHDF5Open (PETSC_COMM_WORLD, "sparse.h5", FILE_MODE_WRITE, &viewer);
+
+  //  PetscErrorCode err = 
+  //  PetscViewerHDF5Open (PETSC_COMM_WORLD, "sparse.h5", FILE_MODE_WRITE, &viewer);
+
+  PetscErrorCode err = 
+    PetscViewerBinaryOpen (PETSC_COMM_WORLD,
+                           "sparse.petsc",
+                           FILE_MODE_WRITE,
+                           &viewer);
+
   MatView (M, viewer);
   PetscViewerDestroy (viewer);
 
diff --git a/src/assign/assign_protein_type.c b/src/assign/assign_protein_type.c
index 065880f..52a0b70 100644
--- a/src/assign/assign_protein_type.c
+++ b/src/assign/assign_protein_type.c
@@ -19,18 +19,26 @@
 /*
  * BLAST database containing all of the influenza protein sequences.
  */
-#define SEQDB "/u/ac/dpellegr/Documents/exp004/influenzadb/influenzadb"
+#define SEQDB "/influenzadb/influenzadb"
 
 /*
  * BLAST reference database of prototypical protein types.
  */
-#define REFDB "/u/ac/dpellegr/Documents/exp004/influenzadb/proteinnames"
+#define REFDB "/influenzadb/proteinnames"
 
 #define BUFFER_LEN 50
 
 void
 assign_protein_type (hid_t file_id)
 {
+  char* exp004 = getenv ("EXP004");
+  if (exp004 == NULL)
+    {
+      printf ("EXP004 environment variable not set.  This should be set to the "
+              "directory containing the genomes/INFLUENZA data.");
+      exit (0);
+    }
+
   /*
    * Iterate through the records for which no protein type has been
    * assigned.  Create a BioSeq Pointer to the data and then use this
@@ -47,7 +55,12 @@ assign_protein_type (hid_t file_id)
   /*
    * Open the BLAST sequence database.
    */
-  ReadDBFILEPtr seqdb = readdb_new (SEQDB, true);
+  char* loc1 = malloc (strlen (exp004) + strlen (SEQDB) + 1);
+  loc1[0] = '\0';
+  strcat (loc1, exp004);
+  strcat (loc1, SEQDB);
+  ReadDBFILEPtr seqdb = readdb_new (loc1, true);
+  free (loc1);
 
   /*
    * Get default BLAST options.
@@ -215,14 +228,19 @@ assign_protein_type (hid_t file_id)
                              "record set.  See the BLAST formatdb.log file "
                              "for details.\n", faa_buf[i].gi);
             }
-
+          
+          char* loc2 = malloc (strlen (exp004) + strlen (REFDB) + 1);
+          loc2[0] = '\0';
+          strcat (loc2, exp004);
+          strcat (loc2, REFDB);
           SeqAlignPtr seqalign = BioseqBlastEngine (bsp,
                                                     "blastp",
-                                                    REFDB,
+                                                    loc2,
                                                     options,
                                                     NULL,
                                                     &error_returns,
                                                     NULL);
+          free (loc2);
 
           /*
            * BLAST reported an error.  Write it out and continue processing.
diff --git a/src/updator.c b/src/updator.c
index 0bade65..b443183 100644
--- a/src/updator.c
+++ b/src/updator.c
@@ -31,9 +31,7 @@ main (int argc, char **argv)
   /*
    * Assign protein type values to the sequence records.
    */
-  /*
   assign_protein_type (file_id);
-  */
 
   /*
    * Assign pairwise BLAST scores.