From 74ae973b5eba7a870c9fe128e422c97c8a778285 Mon Sep 17 00:00:00 2001
From: emmarousseau <emmarou1@icloud.com>
Date: Thu, 23 May 2024 16:21:34 +0200
Subject: [PATCH] Add third test and test data

---
 CHANGELOG.md                                  |  2 +-
 src/umi_tools/umi_tools_dedup/config.vsh.yaml |  2 +-
 src/umi_tools/umi_tools_dedup/script.sh       |  4 +--
 src/umi_tools/umi_tools_dedup/test.sh         | 23 ++++++++++++++
 .../test_data/deduped_unique.sam              | 31 +++++++++++++++++++
 5 files changed, 58 insertions(+), 4 deletions(-)
 create mode 100644 src/umi_tools/umi_tools_dedup/test_data/deduped_unique.sam

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ceb53ac5..bbb4ef4e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -43,7 +43,7 @@
     - `samtools/samtools_stats`: Reports alignment summary statistics for a BAM file (PR #39).
 
 * `umitools`:
-    - `umitools_dedup`: Deduplicate reads based on the mapping co-ordinate and the UMI attached to the read (PR #XXX).
+    - `umitools_dedup`: Deduplicate reads based on the mapping co-ordinate and the UMI attached to the read (PR #54).
 
 ## MAJOR CHANGES
 
diff --git a/src/umi_tools/umi_tools_dedup/config.vsh.yaml b/src/umi_tools/umi_tools_dedup/config.vsh.yaml
index 0c34c20a..cc5113e8 100644
--- a/src/umi_tools/umi_tools_dedup/config.vsh.yaml
+++ b/src/umi_tools/umi_tools_dedup/config.vsh.yaml
@@ -35,7 +35,7 @@ argument_groups:
   - name: Outputs
     arguments:
       - name: --output
-        alternatives: -S
+        alternatives: --stdout
         type: file
         description: Deduplicated BAM file
         required: true
diff --git a/src/umi_tools/umi_tools_dedup/script.sh b/src/umi_tools/umi_tools_dedup/script.sh
index cb5b563a..1e227811 100644
--- a/src/umi_tools/umi_tools_dedup/script.sh
+++ b/src/umi_tools/umi_tools_dedup/script.sh
@@ -24,7 +24,7 @@ test_dir="${metal_executable}/test_data"
 umi_tools dedup \
     --stdin "$par_input" \
     ${par_in_sam:+--in-sam} \
-    -S "$par_output" \
+    --stdout "$par_output" \
     ${par_out_sam:+--out-sam} \
     ${par_paired:+--paired} \
     ${par_output_stats:+--output-stats "$par_output_stats"} \
@@ -52,7 +52,7 @@ umi_tools dedup \
     ${par_mapping_quality:+--mapping-quality "$par_mapping_quality"} \
     ${par_unmapped_reads:+--unmapped-reads "$par_unmapped_reads"} \
     ${par_chimeric_pairs:+--chimeric-pairs "$par_chimeric_pairs"} \
-    ${par_unapired_reads:+--unapired-reads "$par_unapired_reads"} \
+    ${par_unpaired_reads:+--unapired-reads "$par_unapired_reads"} \
     ${par_ignore_umi:+--ignore-umi} \
     ${par_subset:+--subset "$par_subset"} \
     ${par_chrom:+--chrom "$par_chrom"} \
diff --git a/src/umi_tools/umi_tools_dedup/test.sh b/src/umi_tools/umi_tools_dedup/test.sh
index db5563bc..adadb410 100644
--- a/src/umi_tools/umi_tools_dedup/test.sh
+++ b/src/umi_tools/umi_tools_dedup/test.sh
@@ -58,6 +58,29 @@ diff "$out_dir/deduped_fraction.sam" "$test_dir/deduped_fraction.sam" || \
 
 ############################################################################################
 
+echo ">>> Test 3: $meta_functionality_name with --method unique"
+
+"$meta_executable" \
+  --paired \
+  --input "$test_dir/sample.bam" \
+  --bai "$test_dir/sample.bam.bai" \
+  --output "$out_dir/deduped_unique.sam" \
+  --out_sam \
+  --method "unique" \
+  --random_seed 1
+
+echo ">>> Checking whether output exists"
+[ ! -f "$out_dir/deduped_unique.sam" ] && echo "File 'deduped_unique.sam' does not exist!" && exit 1
+
+echo ">>> Checking whether output is non-empty"
+[ ! -s "$out_dir/deduped_unique.sam" ] && echo "File 'deduped_unique.sam' is empty!" && exit 1
+
+echo ">>> Checking whether output is correct"
+diff "$out_dir/deduped_unique.sam" "$test_dir/deduped_unique.sam" || \
+    (echo "Output file deduped_unique.sam does not match expected output" && exit 1)
+
+############################################################################################
+
 rm -rf "$out_dir"
 
 echo "All tests succeeded!"
diff --git a/src/umi_tools/umi_tools_dedup/test_data/deduped_unique.sam b/src/umi_tools/umi_tools_dedup/test_data/deduped_unique.sam
new file mode 100644
index 00000000..570ea153
--- /dev/null
+++ b/src/umi_tools/umi_tools_dedup/test_data/deduped_unique.sam
@@ -0,0 +1,31 @@
+@HD	VN:1.0	SO:coordinate
+@SQ	SN:chr1	LN:197195432
+@SQ	SN:chr10	LN:129993255
+@SQ	SN:chr11	LN:121843856
+@SQ	SN:chr12	LN:121257530
+@SQ	SN:chr13	LN:120284312
+@SQ	SN:chr14	LN:125194864
+@SQ	SN:chr15	LN:103494974
+@SQ	SN:chr16	LN:98319150
+@SQ	SN:chr17	LN:95272651
+@SQ	SN:chr18	LN:90772031
+@SQ	SN:chr19	LN:61342430
+@SQ	SN:chr2	LN:181748087
+@SQ	SN:chr3	LN:159599783
+@SQ	SN:chr4	LN:155630120
+@SQ	SN:chr5	LN:152537259
+@SQ	SN:chr6	LN:149517037
+@SQ	SN:chr7	LN:152524553
+@SQ	SN:chr8	LN:131738871
+@SQ	SN:chr9	LN:124076172
+@SQ	SN:chrM	LN:16299
+@SQ	SN:chrX	LN:166650296
+@SQ	SN:chrY	LN:15902555
+@PG	ID:Bowtie	VN:1.1.2	CL:"bowtie --wrapper basic-0 --threads 4 -v 2 -m 10 -k 1 /ifs/mirror/genomes/bowtie/mm9 /dev/fd/63 --sam"
+@PG	ID:samtools	PN:samtools	PP:Bowtie	VN:1.19.2	CL:samtools view -h example.bam
+@PG	ID:samtools.1	PN:samtools	PP:samtools	VN:1.19.2	CL:samtools view -bS -
+SRR2057595.5052066_ACCGGTTTA	16	chr1	3812794	255	51M	*	0	0	*	*	XA:i:2	MD:Z:42T2T5	NM:i:2
+SRR2057595.13520751_CCAGGTTCT	16	chr1	3967622	255	20M	*	0	0	*	*	XA:i:2	MD:Z:12A0C6	NM:i:2
+SRR2057595.8901432_AGCGGTTAC	0	chr1	4369756	255	20M	*	0	0	*	*	XA:i:2	MD:Z:1T4A13	NM:i:2
+SRR2057595.1210348_ACTGGTTTC	0	chr1	4762503	255	45M	*	0	0	*	*	XA:i:2	MD:Z:0C7A36	NM:i:2
+SRR2057595.1169423_TCTGGTTTC	0	chr1	4762503	255	45M	*	0	0	*	*	XA:i:2	MD:Z:0C7A36	NM:i:2