New command and C functions: obi ecotag

python/obitools3/commands/ecotag.pyx

+#cython: language_level=3
+
+from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
+from obitools3.dms.dms cimport DMS
+from obitools3.dms.view import RollbackException
+from obitools3.dms.capi.obiecotag cimport obi_ecotag
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption
+from obitools3.uri.decode import open_uri
+from obitools3.apps.config import logger
+from obitools3.utils cimport tobytes, str2bytes
+from obitools3.dms.view.view cimport View
+from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
+
+import sys
+
+
+__title__="Taxonomic assignment of sequences"
+
+
+def addOptions(parser):
+
+    addMinimalInputOption(parser)
+    addTaxonomyOption(parser)
+    addMinimalOutputOption(parser)
+ 
+    group = parser.add_argument_group('obi ecotag specific options')
+
+    group.add_argument('--ref-database','-R',
+                      action="store", dest="ecotag:ref_view",
+                      metavar='<REF_VIEW>',
+                      type=str,
+                      help="URI of the view containing the reference database as built by the build_ref_db command.")
+
+    group.add_argument('--minimum-identity','-m',
+                      action="store", dest="ecotag:threshold",
+                      metavar='<THRESHOLD>',
+                      default=0.0,
+                      type=float,
+                      help="Minimum identity to consider for assignment, as a normalized identity, e.g. 0.95 for an identity of 95%%. "
+                           "Default: 0.00 (no threshold).")
+
+def run(config):
+        
+    DMS.obi_atexit()
+    
+    logger("info", "obi ecotag")
+
+    # Open the query view: only the DMS
+    input = open_uri(config['obi']['inputURI'],
+                     dms_only=True)
+    if input is None:
+        raise Exception("Could not read input")
+    i_dms = input[0]
+    i_dms_name = input[0].name
+    i_view_name = input[1]
+
+    # Open the reference view: only the DMS
+    ref = open_uri(config['ecotag']['ref_view'],
+                     dms_only=True)
+    if ref is None:
+        raise Exception("Could not read reference view URI")
+    ref_dms = ref[0]
+    ref_dms_name = ref[0].name
+    ref_view_name = ref[1]
+
+    # Open the output: only the DMS
+    output = open_uri(config['obi']['outputURI'],
+                      input=False,
+                      dms_only=True)
+    if output is None:
+        raise Exception("Could not create output")
+    o_dms = output[0]
+    final_o_view_name = output[1]
+    
+    # If the input and output DMS are not the same, run ecotag creating a temporary view that will be exported to 
+    # the right DMS and deleted in the other afterwards.
+    if i_dms != o_dms:
+        temporary_view_name = final_o_view_name
+        i=0
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = final_o_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+    else:
+        o_view_name = final_o_view_name
+
+    # Read taxonomy DMS and name
+    taxo = open_uri(config['obi']['taxoURI'],
+                    dms_only=True)
+    taxo_dms_name = taxo[0].name
+    taxo_dms = taxo[0]
+    taxonomy_name = config['obi']['taxoURI'].split("/")[-1]   # Robust in theory
+
+    # Save command config in View comments
+    command_line = " ".join(sys.argv[1:])
+    comments = View.print_config(config, "ecotag", command_line, input_dms_name=[i_dms_name], input_view_name=[i_view_name])  # TODO no. fix
+
+    if obi_ecotag(tobytes(i_dms_name), tobytes(i_view_name), \
+                  tobytes(ref_dms_name), tobytes(ref_view_name), \
+                  tobytes(taxo_dms_name), tobytes(taxonomy_name), \
+                  tobytes(o_view_name), comments, 
+                  config['ecotag']['threshold']) < 0:
+        raise Exception("Error running ecotag")
+    
+    # If the input and output DMS are not the same, export result view to output DMS
+    if i_dms != o_dms:
+        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
+    
+    # Save command config in DMS comments
+    o_dms.record_command_line(command_line)
+
+    print("\n")
+    print(repr(o_dms[final_o_view_name]))
+
+    # If the input and the output DMS are different, delete the temporary result view in the input DMS
+    if i_dms != o_dms:
+        View.delete_view(i_dms, o_view_name)
+        o_dms.close()
+    
+    i_dms.close()

python/obitools3/dms/capi/obiecotag.pxd

+#cython: language_level=3
+
+
+cdef extern from "obi_ecotag.h" nogil:
+
+    int obi_ecotag(const char* dms_name,
+                   const char* query_view_name,
+                   const char* ref_dms_name,
+                   const char* ref_view_name,
+                   const char* taxo_dms_name,
+                   const char* taxonomy_name,
+                   const char* output_view_name,
+                   const char* output_view_comments,
+                   double ecotag_threshold)

python/obitools3/dms/dms.cfiles

@@ -38,6 +38,7 @@
 ../../../src/obidmscolumn_str.c
 ../../../src/obidmscolumn.c
 ../../../src/obidmscolumndir.c
+../../../src/obi_ecotag.c
 ../../../src/obierrno.c
 ../../../src/obilittlebigman.c
 ../../../src/obitypes.c

python/obitools3/utils.cfiles

@@ -38,6 +38,7 @@
 ../../src/obidmscolumn_array.c
 ../../src/obidmscolumn.c
 ../../src/obidmscolumndir.c
+../../src/obi_ecotag.c
 ../../src/obierrno.c
 ../../src/obilittlebigman.c
 ../../src/obitypes.c

src/obi_ecotag.h

+/*************************************************************************************************
+ * Header file for functions for the taxonomic assignment of sequences 							 *
+ *************************************************************************************************/
+
+/**
+ * @file obi_ecotag.h
+ * @author Celine Mercier (celine.mercier@metabarcoding.org)
+ * @date November 15th 2018
+ * @brief Header file for the functions for the taxonomic assignment of sequences.
+ */
+
+
+#ifndef OBI_ECOTAG_H_
+#define OBI_ECOTAG_H_
+
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <stdbool.h>
+
+
+
+#define ECOTAG_TAXID_COLUMN_NAME "TAXID"
+#define ECOTAG_NAME_COLUMN_NAME "SCIENTIFIC_NAME"
+#define ECOTAG_STATUS_COLUMN_NAME "ID_STATUS"
+#define ECOTAG_BEST_MATCH_IDS_COLUMN_NAME "BEST_MATCH"
+#define ECOTAG_SCORE_COLUMN_NAME "BEST_IDENTITY"
+
+
+/**
+ * @brief Taxonomic assignment of sequences.
+ *
+ * Note: The columns where the results are written are automatically named and created.
+ *
+ * @param dms_name The path to the DMS where the views are.
+ * @param query_view_name The name of the view containing the query sequences.
+ * @param ref_dms_name The name of the DMS containing the reference database.
+ * @param ref_view_name The name of the view corresponding to the reference database as built by build_reference_db().
+ * @param taxo_dms_name The name of the DMS containing the taxonomy associated with the reference database.
+ * @param taxonomy_name The name of the taxonomy associated with the reference database.
+ * @param output_view_name The name to give to the output view.
+ * @param output_view_comments The comments to associate to the output view.
+ * @param ecotag_threshold The threshold at which to assign.
+ *
+ * 	The algorithm works like this:
+ * 		For each query sequence:
+ *			Align with reference database
+ *			Keep the indices of all the best matches
+ *			For each kept index, get the LCA at that threshold as stored in the reference database, then the LCA of those LCAs
+ *			Write result (max score, threshold, taxid and scientific name of the LCA assigned, list of the ids of the best matches)
+ *
+ * @returns A value indicating the success of the operation.
+ * @retval 0 if the operation was successfully completed.
+ * @retval -1 if an error occurred.
+ *
+ * @since November 2018
+ * @author Celine Mercier (celine.mercier@metabarcoding.org)
+ */
+int obi_ecotag(const char* dms_name,
+		       const char* query_view_name,
+			   const char* ref_dms_name,
+			   const char* ref_view_name,
+			   const char* taxo_dms_name,
+			   const char* taxonomy_name,
+			   const char* output_view_name,
+			   const char* output_view_comments,
+			   double ecotag_threshold);
+
+
+#endif /* OBI_ECOTAG_H_ */
+

src/obierrno.h

@@ -130,6 +130,8 @@ extern int obi_errno;
  	 	 	 	 	 	 	 	 	 	 	 */
 #define OBIVIEW_ALREADY_EXISTS_ERROR (35)   /** Tried to create a new view with a name already existing in the DMS.
 											 */
+#define OBI_ECOTAG_ERROR (36)   			/** Tried to create a new view with a name already existing in the DMS.
+											 */
 /**@}*/
 
 #endif /* OBIERRNO_H_ */