test data files and changes to script

emmarousseau · May 22, 2024 · 13e8703 · 13e8703
1 parent 2e227e4
commit 13e8703
Show file tree

Hide file tree

Showing 6 changed files with 93 additions and 26 deletions.
diff --git a/src/umi_tools/umi_tools_dedup/config.vsh.yaml b/src/umi_tools/umi_tools_dedup/config.vsh.yaml
@@ -27,9 +27,6 @@ argument_groups:
       - name: --bai
         type: file
         description: BAM index
-      - name: --get_output_stats
-        type: boolean_true
-        description: Generate output stats. 
       - name: --random_seed
         type: integer
         description: |
@@ -55,9 +52,10 @@ argument_groups:
           use of the template length to determine reads with the same mapping
           coordinates.
       - name: --output_stats
-        type: file
-        description: Directory containing UMI based deduplication statistics files
-        direction: output
+        type: string
+        description: |
+          Generate files containing UMI based deduplication statistics files with this prefix
+          in the file names.
       - name: --extract_umi_method
         type: string
         description: |
@@ -228,7 +226,7 @@ argument_groups:
         description: |
           Ignore the UMI and group reads using mapping coordinates only.
       - name: --subset
-        type: boolean_true
+        type: double
         description: |
           Only consider a fraction of the reads, chosen at random. This is useful
           for doing saturation analyses.
@@ -269,7 +267,7 @@ argument_groups:
         alternatives: -v
         type: integer
         description: Log level. The higher, the more output.
-        default: 1
+        default: 0
       - name: --error
         alternatives: -E
         type: file

diff --git a/src/umi_tools/umi_tools_dedup/script.sh b/src/umi_tools/umi_tools_dedup/script.sh
@@ -19,13 +19,11 @@ test_dir="${metal_executable}/test_data"
 [[ "$par_ignore_umi" == "false" ]] && unset par_ignore_umi
 [[ "$par_subset" == "false" ]] && unset par_subset
 [[ "$par_log2stderr" == "false" ]] && unset par_log2stderr
-[[ "$par_get_output_stats" == "false" ]] && unset par_get_output_stats
 [[ "$par_read_length" == "false" ]] && unset par_read_length
 
 umi_tools dedup \
     --stdin "$par_input" \
     ${par_in_sam:+--in-sam} \
-    ${par_get_output_stats:+--get-output-stats} \
     -S "$par_output" \
     ${par_out_sam:+--out-sam} \
     ${par_paired:+--paired} \
@@ -56,7 +54,7 @@ umi_tools dedup \
     ${par_chimeric_pairs:+--chimeric-pairs "$par_chimeric_pairs"} \
     ${par_unapired_reads:+--unapired-reads "$par_unapired_reads"} \
     ${par_ignore_umi:+--ignore-umi} \
-    ${par_subset:+--subset} \
+    ${par_subset:+--subset "$par_subset"} \
     ${par_chrom:+--chrom "$par_chrom"} \
     ${par_no_sort_output:+--no-sort-output} \
     ${par_buffer_whole_contig:+--buffer-whole-contig} \

diff --git a/src/umi_tools/umi_tools_dedup/test.sh b/src/umi_tools/umi_tools_dedup/test.sh
@@ -13,39 +13,51 @@ echo ">>> Test 1: Basic usage of $meta_functionality_name with statistics output
   --paired \
   --input "$test_dir/sample.bam" \
   --bai "$test_dir/sample.bam.bai" \
-  --output "$out_dir/deduped.bam" \
-  --output_stats deduped \
+  --output "$out_dir/deduped.sam" \
+  --out_sam \
+  --output_stats "$out_dir/dedup" \
   --random_seed 1
 
 echo ">>> Checking whether output exists"
-[ ! -f "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' does not exist!" && exit 1
-[ ! -f "$out_dir/deduped_edit_distance.tsv" ] && echo "File 'deduped_edit_distance.tsv' does not exist!" && exit 1
+[ ! -f "$out_dir/deduped.sam" ] && echo "File 'deduped.sam' does not exist!" && exit 1
+[ ! -f "$out_dir/dedup_edit_distance.tsv" ] && echo "File 'dedup_edit_distance.tsv' does not exist!" && exit 1
 
 echo ">>> Checking whether output is non-empty"
-[ ! -s "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' is empty!" && exit 1
-[ ! -s "$out_dir/deduped_edit_distance.tsv" ] && echo "File 'deduped_edit_distance.tsv' is empty!" && exit 1
+[ ! -s "$out_dir/deduped.sam" ] && echo "File 'deduped.sam' is empty!" && exit 1
+[ ! -s "$out_dir/dedup_edit_distance.tsv" ] && echo "File 'dedup_edit_distance.tsv' is empty!" && exit 1
 
 echo ">>> Checking whether output is correct"
-diff "$out_dir/deduped.bam" "$test_dir/deduped.bam" || \
-    (echo "Output file deduped.bam does not match expected output" && exit 1)
-diff "$out_dir/deduped_edit_distance.tsv" "$test_dir/deduped_edit_distance.tsv" || \
-    (echo "Output file deduped_edit_distance.tsv does not match expected output" && exit 1)
+diff "$out_dir/deduped.sam" "$test_dir/deduped.sam" || \
+    (echo "Output file deduped.sam does not match expected output" && exit 1)
+diff "$out_dir/dedup_edit_distance.tsv" "$test_dir/dedup_edit_distance.tsv" || \
+    (echo "Output file dedup_edit_distance.tsv does not match expected output" && exit 1)
 
 ############################################################################################
 
-echo ">>> Test 2: $meta_functionality_name"
+echo ">>> Test 2: $meta_functionality_name with random subset selection"
 
 "$meta_executable" \
   --paired \
   --input "$test_dir/sample.bam" \
   --bai "$test_dir/sample.bam.bai" \
-  --output "$out_dir/deduped.bam" \
-  --random_seed 1 \
+  --output "$out_dir/deduped_fraction.sam" \
+  --out_sam \
+  --subset 0.5 \
+  --random_seed 1
 
 
 echo ">>> Checking whether output exists"
-[ ! -f "$out_dir/deduped.bam" ] && echo "File 'deduped.bam' does not exist!" && exit 1
-[]
+[ ! -f "$out_dir/deduped_fraction.sam" ] && echo "File 'deduped_fraction.sam' does not exist!" && exit 1
+
+echo ">>> Checking whether output is non-empty"
+[ ! -s "$out_dir/deduped_fraction.sam" ] && echo "File 'deduped_fraction.sam' is empty!" && exit 1
+
+echo ">>> Checking whether output is correct"
+diff "$out_dir/deduped_fraction.sam" "$test_dir/deduped_fraction.sam" || \
+    (echo "Output file deduped_fraction.sam does not match expected output" && exit 1)
+
+############################################################################################
+
 rm -rf "$out_dir"
 
 echo "All tests succeeded!"

diff --git a/src/umi_tools/umi_tools_dedup/test_data/deduped.bam b/src/umi_tools/umi_tools_dedup/test_data/deduped.bam
diff --git a/src/umi_tools/umi_tools_dedup/test_data/deduped.sam b/src/umi_tools/umi_tools_dedup/test_data/deduped.sam
@@ -0,0 +1,30 @@
+@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
diff --git a/src/umi_tools/umi_tools_dedup/test_data/deduped_fraction.sam b/src/umi_tools/umi_tools_dedup/test_data/deduped_fraction.sam
@@ -0,0 +1,29 @@
+@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 -
+SRR2057595.4062788_ACCGGTTTA	16	chr1	3812793	255	52M	*	0	0	*	*	XA:i:2	MD:Z:43T2T5	NM:i:2
+SRR2057595.8901432_AGCGGTTAC	0	chr1	4369756	255	20M	*	0	0	*	*	XA:i:2	MD:Z:1T4A13	NM:i:2
+SRR2057595.1999468_ACTGGTTTC	0	chr1	4762503	255	45M	*	0	0	*	*	XA:i:2	MD:Z:0C7A36	NM:i:2