From e37ebc5c575696ed7a77f1dcf4521908576ae8b2 Mon Sep 17 00:00:00 2001
From: Lydia Buntrock <lydia.buntrock@fu-berlin.de>
Date: Wed, 28 Jul 2021 13:50:29 +0200
Subject: [PATCH] [FEATURE] Add Tandem Dup to the output vcf

Signed-off-by: Lydia Buntrock <lydia.buntrock@fu-berlin.de>
---
 src/variant_detection/variant_output.cpp | 18 ++++++++++++++++++
 test/data/datasources.cmake              |  2 +-
 test/data/mini_example/output_res.txt    |  1 +
 3 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/src/variant_detection/variant_output.cpp b/src/variant_detection/variant_output.cpp
index 8c2f6569..178a36e7 100644
--- a/src/variant_detection/variant_output.cpp
+++ b/src/variant_detection/variant_output.cpp
@@ -51,6 +51,24 @@ void find_and_output_variants(std::map<std::string, int32_t> & references_length
                             tmp.add_info("END", std::to_string(mate2_pos));
                             tmp.print(out_stream);
                         }
+                        //Tandem Duplication
+                        else if (distance >= args.min_var_length &&
+                                 distance <= args.max_var_length &&
+                                 clusters[i].get_common_tandem_dup_count() > 0 &&
+                                 insert_size >= args.min_var_length)
+                        {
+                            variant_record tmp{};
+                            tmp.set_chrom(mate1.seq_name);
+                            tmp.set_qual(cluster_size);
+                            tmp.set_alt("<DUP:TANDEM>");
+                            tmp.add_info("SVTYPE", "DUP");
+                            // Increment position by 1 because VCF is 1-based
+                            tmp.set_pos(mate1_pos + 1);
+                            tmp.add_info("SVLEN", std::to_string(distance));
+                            // Increment end by 1 because VCF is 1-based
+                            tmp.add_info("END", std::to_string(mate2_pos + 1));
+                            tmp.print(out_stream);
+                        }
                         //Insertion
                         else if (distance == 1 &&
                                 insert_size >= args.min_var_length)
diff --git a/test/data/datasources.cmake b/test/data/datasources.cmake
index 204f8ea3..32e1b5ff 100644
--- a/test/data/datasources.cmake
+++ b/test/data/datasources.cmake
@@ -29,4 +29,4 @@ declare_datasource (FILE output_err.txt
 # copies file to <build>/data/output_res.txt
 declare_datasource (FILE output_res.txt
                     URL ${CMAKE_SOURCE_DIR}/test/data/mini_example/output_res.txt
-                    URL_HASH SHA256=7f52103b216002598f8a50bd3f27f430798de7ffc03ae26c92bf3281ed338164)
+                    URL_HASH SHA256=81f861156dfdff027df053e996a1dfaeccf15131ce81c5ccb350148fc912a8c7)
diff --git a/test/data/mini_example/output_res.txt b/test/data/mini_example/output_res.txt
index f9771cb0..5844eb14 100644
--- a/test/data/mini_example/output_res.txt
+++ b/test/data/mini_example/output_res.txt
@@ -12,6 +12,7 @@
 chr1	57	.	N	<DEL>	9	PASS	END=70;SVLEN=-13;SVTYPE=DEL	GT	./.
 chr1	97	.	N	<DEL>	1	PASS	END=125;SVLEN=-28;SVTYPE=DEL	GT	./.
 chr1	125	.	N	<INS>	3	PASS	END=125;SVLEN=15;SVTYPE=INS	GT	./.
+chr1	180	.	N	<DUP:TANDEM>	1	PASS	END=188;SVLEN=8;SVTYPE=DUP	GT	./.
 chr1	266	.	N	<DEL>	4	PASS	END=286;SVLEN=-20;SVTYPE=DEL	GT	./.
 chr1	282	.	N	<DEL>	1	PASS	END=299;SVLEN=-17;SVTYPE=DEL	GT	./.
 chr1	336	.	N	<DEL>	4	PASS	END=350;SVLEN=-14;SVTYPE=DEL	GT	./.