build_ref_db: fixed bug erasing some of the higher LCAs (i.e. lowest

similarities)

python/obitools3/version.py

@@ -1,5 +1,5 @@
 major = 3
 minor = 0
-serial= '0-beta14e'
+serial= '0-beta15'
 
 version ="%d.%02d.%s" % (major,minor,serial)

src/build_reference_db.c

@@ -477,6 +477,8 @@ int build_reference_db(const char* dms_name,
 		// Read alignment score
 		score = obi_get_float_with_elt_idx_and_col_p_in_view(matrix_with_lca_view, matrix_score_column, i, 0);
 
+		//fprintf(stderr, "\n\ntaxid_lca=%d, score=%f, idx1=%d, idx2=%d", taxid_lca, score, idx1, idx2);
+
 		///////////////// Compute for first sequence \\\\\\\\\\\\\\\\\\\\\\\     (TODO function)
 
 		// Read arrays
@@ -493,9 +495,11 @@ int build_reference_db(const char* dms_name,
 //			return -1;
 //		}
 
+		//fprintf(stderr, "\n1st sequence");
 		// If empty, add values
 		if (taxid_array_length == 0)
 		{
+			//fprintf(stderr, "\nEmpty, add value");
 			if (obi_set_array_with_col_p_in_view(o_view, final_lca_taxid_a_column, idx1, &taxid_lca, (uint8_t) (obi_sizeof(OBI_INT) * 8), 1) < 0)
 			{
 				obidebug(1, "\nError setting a LCA taxid array in a column when building a reference database");
@@ -509,6 +513,8 @@ int build_reference_db(const char* dms_name,
 		}
 		else
 		{
+			//fprintf(stderr, "\nNot empty");
+
 			j = 0;
 			modified = false;
 			while (j < taxid_array_length)
@@ -522,6 +528,9 @@ int build_reference_db(const char* dms_name,
 						memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 						modified = true;
 
+						//fprintf(stderr, "\nSame LCA, replace %d and %f with %d and %f", lca_taxid_array_writable[j],
+						//		score_array_writable[j], taxid_lca, score);
+
 						// Better score for the same LCA, replace this LCA/score pair
 						lca_taxid_array_writable[j] = taxid_lca;
 						score_array_writable[j] = score;
@@ -548,6 +557,8 @@ int build_reference_db(const char* dms_name,
 				{
 					if (score > score_array[j])
 					{
+						//fprintf(stderr, "\nInsert new");
+
 						memcpy(lca_taxid_array_writable, lca_taxid_array, taxid_array_length*sizeof(obiint_t));
 						memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 						modified = true;
@@ -592,10 +603,15 @@ int build_reference_db(const char* dms_name,
 				memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 				modified = true;
 
+				//fprintf(stderr, "\nAppend at the end");
+
 				// Append LCA
 				lca_taxid_array_writable[taxid_array_writable_length] = taxid_lca;
 				score_array_writable[score_array_writable_length] = score;
 
+				taxid_array_writable_length++;
+				score_array_writable_length++;
+
 				// Remove the previous (children) LCAs from the array if their score is equal or lower
 				while ((j>0) && (score_array_writable[j-1] <= score))
 				{
@@ -616,6 +632,13 @@ int build_reference_db(const char* dms_name,
 			// Write new arrays
 			if (modified)
 			{
+//				fprintf(stderr, "\n\nnew array:");
+//				for (k=0;k<taxid_array_writable_length;k++)
+//				{
+//					lca = obi_taxo_get_taxon_with_taxid(tax, lca_taxid_array_writable[k]);
+//					fprintf(stderr, "\nLCA=%d, %s, score=%f", lca_taxid_array_writable[k], lca->name, score_array_writable[k]);
+//				}
+
 				if (obi_set_array_with_col_p_in_view(o_view, final_lca_taxid_a_column, idx1, lca_taxid_array_writable, (uint8_t) (obi_sizeof(OBI_INT) * 8), taxid_array_writable_length) < 0)
 				{
 					obidebug(1, "\nError setting a LCA taxid array in a column when building a reference database");
@@ -645,9 +668,13 @@ int build_reference_db(const char* dms_name,
 //			return -1;
 //		}
 
+		//fprintf(stderr, "\n2nd sequence");
+
 		// If empty, add values
 		if (taxid_array_length == 0)
 		{
+			//fprintf(stderr, "\nEmpty, add value");
+
 			if (obi_set_array_with_col_p_in_view(o_view, final_lca_taxid_a_column, idx2, &taxid_lca, (uint8_t) (obi_sizeof(OBI_INT) * 8), 1) < 0)
 			{
 				obidebug(1, "\nError setting a LCA taxid array in a column when building a reference database");
@@ -661,6 +688,8 @@ int build_reference_db(const char* dms_name,
 		}
 		else
 		{
+			//fprintf(stderr, "\nNot empty");
+
 			j = 0;
 			modified = false;
 			while (j < taxid_array_length)
@@ -674,6 +703,9 @@ int build_reference_db(const char* dms_name,
 						memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 						modified = true;
 
+						//fprintf(stderr, "\nSame LCA, replace %d and %f with %d and %f", lca_taxid_array_writable[j],
+						//		score_array_writable[j], taxid_lca, score);
+
 						// Better score for the same LCA, replace this LCA/score pair
 						lca_taxid_array_writable[j] = taxid_lca;
 						score_array_writable[j] = score;
@@ -700,6 +732,8 @@ int build_reference_db(const char* dms_name,
 				{
 					if (score > score_array[j])
 					{
+						//fprintf(stderr, "\nInsert new");
+
 						memcpy(lca_taxid_array_writable, lca_taxid_array, taxid_array_length*sizeof(obiint_t));
 						memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 						modified = true;
@@ -740,6 +774,8 @@ int build_reference_db(const char* dms_name,
 
 			if (j == taxid_array_length) // same or parent LCA not found, need to be appended at the end
 			{
+				//fprintf(stderr, "\nAppend at the end");
+
 				memcpy(lca_taxid_array_writable, lca_taxid_array, taxid_array_length*sizeof(obiint_t));
 				memcpy(score_array_writable, score_array, score_array_length*sizeof(obifloat_t));
 				modified = true;
@@ -748,6 +784,9 @@ int build_reference_db(const char* dms_name,
 				lca_taxid_array_writable[taxid_array_writable_length] = taxid_lca;
 				score_array_writable[score_array_writable_length] = score;
 
+				taxid_array_writable_length++;
+				score_array_writable_length++;
+
 				// Remove the previous (children) LCAs from the array if their score is equal or lower
 				while ((j>0) && (score_array_writable[j-1] <= score))
 				{