ecotag: BEST_MATCH_TAXIDS now dereplicated (no repeated taxids in the

list) and switch to version 3.0.1b8

python/obitools3/version.py

@@ -1,5 +1,5 @@
 major = 3
 minor = 0
-serial= '1b7'
+serial= '1b8'
 
 version ="%d.%d.%s" % (major,minor,serial)

src/obi_ecotag.c

@@ -233,7 +233,7 @@ int obi_ecotag(const char* dms_name,
 		// Write result (max score, threshold, LCA assigned, list of the ids of the best matches)
 
 
-	index_t         i, j, k;
+	index_t         i, j, k, t;
 	ecotx_t*        lca;
 	ecotx_t*        lca_in_array;
 	ecotx_t*        best_match;
@@ -259,16 +259,20 @@ int obi_ecotag(const char* dms_name,
 	int32_t*   		best_match_taxids;
 	int32_t*		best_match_taxids_to_store;
 	int				best_match_count;
+	int				best_match_taxid_count;
 	int             buffer_size;
 	int 			best_match_ids_buffer_size;
 	index_t         best_match_idx;
 	int32_t         lca_array_length;
 	int32_t         lca_taxid;
 	int32_t         taxid_best_match;
+	int32_t         taxid;
+	int32_t         taxid_to_store;
 	bool   			assigned;
 	const char*		lca_name;
 	const char*		id;
 	int				id_len;
+	bool			already_in;
 
 	OBIDMS_p          dms = NULL;
 	OBIDMS_p		  ref_dms = NULL;
@@ -488,10 +492,11 @@ int obi_ecotag(const char* dms_name,
 
 	for (i=0; i < query_count; i++)
 	{
-		if (i%1000 == 0)
+		if (i%10 == 0)
 			fprintf(stderr,"\rDone : %f %%       ", (i / (float) query_count)*100);
 
 		best_match_count = 0;
+		best_match_taxid_count = 0;
 		best_match_ids_length = 0;
 		threshold = ecotag_threshold;
 		best_score = 0.0;
@@ -543,6 +548,7 @@ int obi_ecotag(const char* dms_name,
 					// Reset the array with that match
 					best_match_ids_length = 0;
 					best_match_count = 0;
+					best_match_taxid_count = 0;
 				}
 
 				// Store in best match array
@@ -585,8 +591,27 @@ int obi_ecotag(const char* dms_name,
 
 				// Save match
 				best_match_array[best_match_count] = j;
-				best_match_taxids[best_match_count] = obi_get_int_with_elt_idx_and_col_p_in_view(ref_view, ref_taxid_column, j, 0);
 				best_match_count++;
+
+				// Save best match taxid only if not already in array
+				taxid_to_store = obi_get_int_with_elt_idx_and_col_p_in_view(ref_view, ref_taxid_column, j, 0);
+				already_in = false;
+				for (t=0; t<best_match_taxid_count; t++)
+				{
+					taxid = best_match_taxids[t];
+					//fprintf(stderr, "\ntaxid %d, taxid_to_store %d\n", taxid, taxid_to_store);
+					if (taxid == taxid_to_store)
+					{
+						already_in = true;
+						break;
+					}
+				}
+				if (! already_in)
+				{
+					best_match_taxids[best_match_taxid_count] = taxid_to_store;
+					best_match_taxid_count++;
+				}
+
 				strcpy(best_match_ids+best_match_ids_length, id);
 				best_match_ids_length = best_match_ids_length + id_len + 1;
 			}
@@ -693,7 +718,7 @@ int obi_ecotag(const char* dms_name,
 									assigned_name_column, lca_name,
 									assigned_status_column, assigned,
 									best_match_ids_column, best_match_ids_to_store, best_match_ids_length,
-									best_match_taxids_column, best_match_taxids_to_store, best_match_count,
+									best_match_taxids_column, best_match_taxids_to_store, best_match_taxid_count,
 									score_column, best_score
 									) < 0)
 							return -1;