From 946f9723b877154f4c60ebbb3ba8da3e80fa66a2 Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Tue, 23 Jul 2019 19:02:17 +0200
Subject: [PATCH] ecotag: fixed a bug where the wrong taxid for the best match
 was retrieved

---
 src/obi_ecotag.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/src/obi_ecotag.c b/src/obi_ecotag.c
index ca745e7..9f08941 100755
--- a/src/obi_ecotag.c
+++ b/src/obi_ecotag.c
@@ -394,8 +394,7 @@ int obi_ecotag(const char* dms_name,
 	}
 
 	// Check if the sequence columns share the same indexer (allows for quick checking of sequence equality)
-	// TODO check same dms
-	if (strcmp((query_seq_column->header)->indexer_name, (ref_seq_column->header)->indexer_name) == 0)
+	if ((strcmp((query_seq_column->header)->indexer_name, (ref_seq_column->header)->indexer_name) == 0) && (dms == ref_dms))
 		same_indexer = true;
 	else
 		same_indexer = false;
@@ -466,12 +465,9 @@ int obi_ecotag(const char* dms_name,
 			{
 				// kmer filter (offset for the index of the kmer table of the 2nd sequence because the kmer tables of both the query and ref columns are concatenated in one)
 				align_filters(ktable, blob1, blob2, i, query_count+j, threshold, true, ALILEN, true, &score, &lcs_min, !same_indexer);
-
 				// Compute alignment score
 				if ((score < 0) && ((threshold == 0) || (score == -1.0)))	// (sequences are not identical), and (no threshold, or filter passed): align
-				{
 					score = obiblob_sse_banded_lcs_align(blob1, blob2, threshold, true, ALILEN, true, &lcs_length, &ali_length);
-				}
 			}
 
 			if ((score >= threshold) && (score >= best_score))
@@ -527,6 +523,8 @@ int obi_ecotag(const char* dms_name,
 		}
 
 		// Get LCA of the LCAs of the best matches
+		lca = NULL;
+		lca_taxid = -1;
 		for (j=0; j<best_match_count; j++)
 		{
 			best_match_idx = best_match_array[j];
@@ -535,7 +533,6 @@ int obi_ecotag(const char* dms_name,
 			score_array = obi_get_array_with_col_p_in_view(ref_view, score_a_column, best_match_idx, &lca_array_length);
 
 			k = 0;
-
 			while ((k < lca_array_length) && (score_array[k] >= ecotag_threshold))
 				k++;
 
@@ -572,7 +569,7 @@ int obi_ecotag(const char* dms_name,
 			}
 			else
 			{
-				taxid_best_match = obi_get_int_with_elt_idx_and_col_p_in_view(ref_view, ref_taxid_column, j, 0);
+				taxid_best_match = obi_get_int_with_elt_idx_and_col_p_in_view(ref_view, ref_taxid_column, best_match_idx, 0);
 				best_match = obi_taxo_get_taxon_with_taxid(taxonomy, taxid_best_match);
 				if (best_match == NULL)
 				{
@@ -592,6 +589,7 @@ int obi_ecotag(const char* dms_name,
 			}
 		}
 
+		lca_name = NULL;
 		if (best_match_count > 0)
 		{
 			assigned = true;