ecotag: added separate threshold for minimum circle identity (and switch

to version 3.0.0b35

python/obitools3/commands/cat.pyx

@@ -48,9 +48,9 @@ def run(config):
     
     logger("info", "obi cat")
 
-    # Open the views to concatenate
-    iview_list = []
+    # Check the views to concatenate
     idms_list = []
+    iview_list = []
     total_len = 0
     remove_qual = False
     remove_rev_qual = False
@@ -68,8 +68,9 @@ def run(config):
         if REVERSE_QUALITY_COLUMN not in i_view: # same as above for reverse quality
             remove_rev_qual = True
         total_len += len(i_view)
-        iview_list.append(i_view)
         idms_list.append(i_dms)
+        iview_list.append(i_view.name)
+        i_view.close()
 
     # Open the output: only the DMS
     output = open_uri(config['obi']['outputURI'],
@@ -98,8 +99,10 @@ def run(config):
     # Initialize multiple elements columns
     if type(output_0)==BufferedWriter: 
         dict_cols = {}
-        for v in iview_list:
+        for v_uri in config["cat"]["views_to_cat"]:
+            v = open_uri(v_uri)[1]
             for coln in v.keys():
+                col = v[coln]
                 if v[coln].nb_elements_per_line > 1:
                     if coln not in dict_cols:
                         dict_cols[coln] = {}
@@ -109,6 +112,7 @@ def run(config):
                     else:
                         dict_cols[coln]['eltnames'] = set(v[coln].elements_names + list(dict_cols[coln]['eltnames']))
                         dict_cols[coln]['nbelts'] = len(dict_cols[coln]['eltnames'])
+            v.close()            
         for coln in dict_cols:
             Column.new_column(o_view, coln, dict_cols[coln]['obitype'], 
                               nb_elements_per_line=dict_cols[coln]['nbelts'], elements_names=list(dict_cols[coln]['eltnames']))
@@ -120,7 +124,8 @@ def run(config):
         pb = None
     
     i = 0
-    for v in iview_list:
+    for v_uri in config["cat"]["views_to_cat"]:
+        v = open_uri(v_uri)[1]
         for entry in v:
             PyErr_CheckSignals()
             if pb is not None:
@@ -131,6 +136,7 @@ def run(config):
             else:
                 o_view[i] = entry
             i+=1
+        v.close()
 
     # Deletes quality columns if needed
     if type(output_0)!=BufferedWriter: 
@@ -145,7 +151,7 @@ def run(config):
     
     # Save command config in DMS comments
     command_line = " ".join(sys.argv[1:])
-    o_view.write_config(config, "cat", command_line, input_dms_name=[d.name for d in idms_list], input_view_name=[v.name for v in iview_list])
+    o_view.write_config(config, "cat", command_line, input_dms_name=[d.name for d in idms_list], input_view_name=[vname for vname in iview_list])
     o_dms.record_command_line(command_line)
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)

python/obitools3/commands/ecotag.pyx

@@ -41,6 +41,17 @@ def addOptions(parser):
                       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).")
 
+    group.add_argument('--minimum-circle','-c',
+                      action="store", dest="ecotag:bubble_threshold",
+                      metavar='<CIRCLE_THRESHOLD>',
+                      default=0.99,
+                      type=float,
+                      help="Minimum identity considered for the assignment circle "
+                           "(sequence is assigned to the LCA of all sequences within a similarity circle of the best matches; "
+                           "the threshold for this circle is the highest value between <CIRCLE_THRESHOLD> and the best assignment score found). "
+                           "Give value as a normalized identity, e.g. 0.95 for an identity of 95%%. "
+                           "Default: 0.99.")
+
 def run(config):
         
     DMS.obi_atexit()
@@ -66,9 +77,8 @@ def run(config):
     ref_view_name = ref[1]
 
     # Check that the threshold demanded is greater than or equal to the threshold used to build the reference database
-    if config['ecotag']['threshold'] < eval(ref_dms[ref_view_name].comments["ref_db_threshold"]) :
-        print("Error: The threshold demanded (%f) is lower than the threshold used to build the reference database (%f).", 
-              config['ecotag']['threshold'], ref_dms[ref_view_name].comments["ref_db_threshold"])
+    if config['ecotag']['bubble_threshold'] < eval(ref_dms[ref_view_name].comments["ref_db_threshold"]) :
+        raise Exception(f"Error: The threshold demanded ({config['ecotag']['bubble_threshold']}) is lower than the threshold used to build the reference database ({float(ref_dms[ref_view_name].comments['ref_db_threshold'])}).")
     
     # Open the output: only the DMS
     output = open_uri(config['obi']['outputURI'],
@@ -113,8 +123,9 @@ def run(config):
     if obi_ecotag(i_dms.name_with_full_path, tobytes(i_view_name), \
                   ref_dms.name_with_full_path, tobytes(ref_view_name), \
                   taxo_dms.name_with_full_path, tobytes(taxonomy_name), \
-                  tobytes(o_view_name), comments, 
-                  config['ecotag']['threshold']) < 0:
+                  tobytes(o_view_name), comments, \
+                  config['ecotag']['threshold'], \
+                  config['ecotag']['bubble_threshold']) < 0:
         raise Exception("Error running ecotag")
     
     # If the input and output DMS are not the same, export result view to output DMS

python/obitools3/dms/capi/obiecotag.pxd

@@ -11,4 +11,5 @@ cdef extern from "obi_ecotag.h" nogil:
                    const char* taxonomy_name,
                    const char* output_view_name,
                    const char* output_view_comments,
-                   double ecotag_threshold)
+                   double ecotag_threshold,
+                   double bubble_threshold)

python/obitools3/version.py

@@ -1,5 +1,5 @@
 major = 3
 minor = 0
-serial= '0b33'
+serial= '0b35'
 
 version ="%d.%d.%s" % (major,minor,serial)

src/obi_ecotag.c

@@ -218,7 +218,8 @@ int obi_ecotag(const char* dms_name,
 			   const char* taxonomy_name,
 			   const char* output_view_name,
 			   const char* output_view_comments,
-			   double ecotag_threshold)    // TODO different threshold for the similarity sphere around ref seqs
+			   double ecotag_threshold,
+			   double bubble_threshold)
 {
 
 	// For each sequence
@@ -239,6 +240,7 @@ int obi_ecotag(const char* dms_name,
 	index_t         query_seq_idx, ref_seq_idx;
 	double          score, best_score;
 	double			threshold;
+	double			lca_threshold;
 	int             lcs_length;
 	int             ali_length;
 	Kmer_table_p    ktable;
@@ -389,10 +391,10 @@ int obi_ecotag(const char* dms_name,
 		return -1;
 	}
 	free(db_threshold_str);
-	if (ecotag_threshold < db_threshold)
+	if (bubble_threshold < db_threshold)
 	{
 		fprintf(stderr, "\nError: The threshold demanded (%f) is lower than the threshold used to build the reference database (%f).\n\n",
-				ecotag_threshold, db_threshold);
+				bubble_threshold, db_threshold);
 		return -1;
 	}
 
@@ -597,11 +599,16 @@ int obi_ecotag(const char* dms_name,
 		{
 			best_match_idx = best_match_array[j];
 
-			// Find the LCA for the chosen threshold
+			// Find the LCA for the highest threshold between best_score and the chosen bubble threshold
 			score_array = obi_get_array_with_col_p_in_view(ref_view, score_a_column, best_match_idx, &lca_array_length);
 
+			if (bubble_threshold < best_score)
+				lca_threshold = best_score;
+			else
+				lca_threshold = bubble_threshold;
+
 			k = 0;
-			while ((k < lca_array_length) && (score_array[k] >= best_score))
+			while ((k < lca_array_length) && (score_array[k] >= lca_threshold))
 				k++;
 
 			if (k>0)

src/obi_ecotag.h

@@ -42,12 +42,14 @@
  * @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.
+ * @param bubble_threshold The threshold at which to look for an LCA (i.e. minimum identity considered for the assignment circle);
+ *                         the threshold actually used will be the highest between this value and the best assignment score found.
  *
  * 	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
+ *			For each kept index, get the LCA at the highest threshold between bubble_threshold and the best assignment score found (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.
@@ -65,7 +67,8 @@ int obi_ecotag(const char* dms_name,
 			   const char* taxonomy_name,
 			   const char* output_view_name,
 			   const char* output_view_comments,
-			   double ecotag_threshold);
+			   double ecotag_threshold,
+			   double bubble_threshold);
 
 
 #endif /* OBI_ECOTAG_H_ */