22 files changed, 181 insertions, 71 deletions

diff --git a/src/assign/assign_protein_type.h b/src/assign/assign_protein_type.h
new file mode 100644
index 0000000..221154f
--- a/dev/null
+++ b/src/assign/assign_protein_type.h
@@ -0,0 +1,15 @@
+#ifndef ASSIGN_PROTEIN_TYPE_H
+#define ASSIGN_PROTEIN_TYPE_H
+
+#include <hdf5.h>
+
+/*
+ * Determine the protein type for each protein sequence record.  The
+ * technique used by NCBI is used here.  A BLAST database of
+ * prototypical protein sequences serves as the reference.  Each input
+ * sequence is BLASTed against this database.  The first hit is used
+ * to assign a protein type to sequence.
+ */
+void assign_protein_type (hid_t file_id);
+
+#endif // ASSIGN_PROTEIN_TYPE_H