Add tests and adjust script

src/samtools/samtools_faidx/config.vsh.yaml

@@ -4,7 +4,7 @@ description: Indexes FASTA files to enable random access to fasta and fastq file
 keywords: [ idex, fasta, faidx ]
 links:
   homepage: https://www.htslib.org/
-  documentation: https://www.htslib.org/doc/faidx.html
+  documentation: https://www.htslib.org/doc/samtools-faidx.html
   repository: https://github.com/samtools/samtools
 references: 
   doi: [10.1093/bioinformatics/btp352, 10.1093/gigascience/giab008]
@@ -13,27 +13,29 @@ license: MIT/Expat
 argument_groups:
   - name: Inputs
     arguments:
-    - name: --fasta
+    - name: --input
       type: file
       description: |
         FASTA input file.
     - name: --length 
       alternatives: -n
       type: integer
       description: |
-        Length of FASTA sequence line.
+        Length for FASTA sequence line wrapping. If zero, this means do not
+        line wrap. Defaults to the line length in the input file.
       default: 60
     - name: --region_file
       alternatives: -r
       type: file
       description: |
-        File of regions.  Format is chr:from-to. One per line.
+        File of regions. Format is chr:from-to. One per line.
+        Must be used with --output to avoid sending output to stdout.
   - name: Options
     arguments:
     - name: --continue
       type: boolean_true
       description: |
-        Continue after trying to retrieve missing region.
+        Continue working if a non-existent region is requested.
     - name: --reverse_complement
       alternatives: -i
       type: boolean_true
@@ -45,9 +47,9 @@ argument_groups:
       alternatives: -o
       type: file 
       description: |
-        Write FASTA to file.
-      required: true
+        Write output to file.
       direction: output
+      required: true
       example: output.fasta
     - name: --mark_strand
       type: string
@@ -58,19 +60,19 @@ argument_groups:
     - name: --fai_idx
       type: file
       description: |
-        Name of the index file (default file.fa.fai).
+        Read/Write to specified index file (default file.fa.fai).
       direction: output
       example: file.fa.fai
     - name: --gzi_idx
       type: file
       description: |
-        Name of compressed file index (default file.fa.gz.gzi).
+        Read/Write to specified compressed file index (used with .gz files, default file.fa.gz.gzi).
       direction: output
       example: file.fa.gz.gzi
     - name: --fastq
       type: boolean_true
       description: |
-        File and index in FASTQ format.
+        Read FASTQ files and output extracted sequences in FASTQ format. Same as using samtools fqidx.
    
 resources:
   - type: bash_script

src/samtools/samtools_faidx/script.sh

@@ -10,12 +10,12 @@ set -e
 [[ "$par_fastq" == "false" ]] && unset par_fastq
 
 samtools faidx \
-    "$par_fasta" \
-    -o "$par_output" \
-    ${par_length:+-l "$par_length"} \
+    "$par_input" \
+    ${par_output:+-o "$par_output"} \
+    ${par_length:+-n "$par_length"} \
     ${par_continue:+-c} \
-    ${part_region_file:+-r "$par_region_file"} \
-    ${par_revferse_complement:+-r} \
+    ${par_region_file:+-r "$par_region_file"} \
+    ${par_reverse_complement:+-r} \
     ${par_mark_strand:+--mark-strand "$par_mark_strand"} \
     ${par_fai_idx:+--fai-idx "$par_fai_idx"} \
     ${par_gzi_idx:+--gzi-idx "$par_gzi_idx"} \

src/samtools/samtools_faidx/test.sh

@@ -4,36 +4,101 @@ test_dir="${meta_resources_dir}/test_data"
 echo ">>> Testing $meta_functionality_name"
 
 "$meta_executable" \
-  "$test_dir/test.fasta" \
-  --output "$test_dir/test.fasta.fai" \
-  --continue \
-  --fai-idx
+  --input "$test_dir/test.fasta" \
+  --output "$test_dir/test.fasta.fai"
+
+echo "$meta_executable"
+echo "$test_dir/test.fasta"
 
 echo ">>> Checking whether output exists"
 [ ! -f "$test_dir/test.fasta.fai" ] && echo "File 'test.fasta.fai' does not exist!" && exit 1
-[ ! -f "$test_dir/test.fasta.fai.idx" ] && echo "File 'test.fasta.fai.idx' does not exist!" && exit 1
-[ ! -f "$test_dir/test.fasta.gzi.idx" ] && echo "File 'test.fasta.gzi.idx' does not exist!" && exit 1
 
 echo ">>> Checking whether output is non-empty"
 [ ! -s "$test_dir/test.fasta.fai" ] && echo "File 'test.fasta.fai' is empty!" && exit 1
-[ ! -s "$test_dir/test.fasta.fai.idx" ] && echo "File 'test.fasta.fai.idx' is empty!" && exit 1
-[ ! -s "$test_dir/test.fasta.gzi.idx" ] && echo "File 'test.fasta.gzi.idx' is empty!" && exit 1
 
 echo ">>> Checking whether output is correct"
-diff "$test_dir/a.flagstat" "$test_dir/a_ref.flagstat" || \
-    (echo "Output file a.flagstat does not match expected output" && exit 1)
+diff "$test_dir/test.fasta.fai" "$test_dir/output/test.fasta.fai" || \
+    (echo "Output file test.fasta.fai does not match expected output" && exit 1)
+
+rm "$test_dir/test.fasta.fai"
+
+####################################################################################################
+
+echo ">>> Test 2: ${meta_functionality_name} with bgzipped input"
+
+"$meta_executable" \
+  --input "$test_dir/test.fasta.gz" \
+  --output "$test_dir/test.fasta.gz.fai"
+
+echo ">>> Checking whether output exists"1
+[ ! -f "$test_dir/test.fasta.gz.fai" ] && echo "File 'test.fasta.gz.fai' does not exist!" && exit 1
+[ ! -f "$test_dir/test.fasta.gz.gzi" ] && echo "File 'test.fasta.gz.gzi' does not exist!" && exit 1
+
+echo ">>> Checking whether output is non-empty"
+[ ! -s "$test_dir/test.fasta.gz.fai" ] && echo "File 'test.fasta.gz.fai' is empty!" && exit 1
+[ ! -s "$test_dir/test.fasta.gz.gzi" ] && echo "File 'test.fasta.gz.gzi' is empty!" && exit 1
+
+echo ">>> Checking whether output is correct"
+diff "$test_dir/test.fasta.gz.fai" "$test_dir/output/test.fasta.gz.fai" || \
+    (echo "Output file test_zip.fasta.gz.fai does not match expected output" && exit 1)
+diff "$test_dir/test.fasta.gz.gzi" "$test_dir/output/test.fasta.gz.gzi" || \
+    (echo "Output file test2.fasta.gz.gzi does not match expected output" && exit 1)
+
+rm "$test_dir/test.fasta.gz.fai"
+rm "$test_dir/test.fasta.gz.gzi"
+
+####################################################################################################
+
+echo ">>> Test 3: ${meta_functionality_name} with fastq input"
+
+"$meta_executable" \
+  --input "$test_dir/test.fastq" \
+  --output "$test_dir/test.fastq.fai"
+
+echo ">>> Checking whether output exists"
+[ ! -f "$test_dir/test.fastq.fai" ] && echo "File 'test.fastq.fai' does not exist!" && exit 1
+
+echo ">>> Checking whether output is non-empty"
+[ ! -s "$test_dir/test.fastq.fai" ] && echo "File 'test.fastq.fai' is empty!" && exit 1
+
+echo ">>> Checking whether output is correct"
+diff "$test_dir/test.fastq.fai" "$test_dir/output/test.fastq.fai" || \
+    (echo "Output file test.fastq.fai does not match expected output" && exit 1)
+
+rm "$test_dir/test.fastq.fai"
 
-rm "$test_dir/a.flagstat"
+####################################################################################################
 
-echo ">>> Test 2:"
+echo ">>> Test 4: ${meta_functionality_name} with region file containing non-existent regions and 
+      specific fasta line wrap length"
 
 "$meta_executable" \
-  "$test_dir/test.fasta" \
-  --output "$test_dir/test.fasta.fai" \
-  --length 60 \
+  --input "$test_dir/test.fasta" \
+  --output "$test_dir/regions.fasta" \
+  --length 10 \
   --continue \
-  --gzi-idx "$test_dir/test.fasta.gz.gzi" \
+  --region_file "$test_dir/test.regions" \
+  --fai_idx "$test_dir/regions.fasta.fai"
 
+echo ">>> Checking whether output exists"
+[ ! -f "$test_dir/regions.fasta" ] && echo "File 'regions.fasta' does not exist!" && exit 1
+[ ! -f "$test_dir/regions.fasta.fai" ] && echo "File 'regions.fasta.fai' does not exist!" && exit 1
+
+echo ">>> Checking whether output is non-empty"
+[ ! -s "$test_dir/regions.fasta" ] && echo "File 'regions.fasta' is empty!" && exit 1
+[ ! -s "$test_dir/regions.fasta.fai" ] && echo "File 'regions.fasta.fai' is empty!" && exit 1
+
+echo ">>> Checking whether output is correct"
+diff "$test_dir/regions.fasta" "$test_dir/output/regions.fasta" || \
+    (echo "Output file regions.fasta does not match expected output" && exit 1)
+diff "$test_dir/regions.fasta.fai" "$test_dir/output/regions.fasta.fai" || \
+    (echo "Output file regions.fasta.fai does not match expected output" && exit 1)
+
+rm "$test_dir/regions.fasta"
+rm "$test_dir/regions.fasta.fai"
+
+####################################################################################################
 
 echo "All tests succeeded!"
-exit 0
+exit 0
+

src/samtools/samtools_faidx/test_data/output/regions.fasta

@@ -0,0 +1,14 @@
+>YAL069W:300-315
+CCCAAATATT
+GTATAA
+>YAL068C:200-230
+CTGAAGCCGT
+TTTCAACTAC
+GGTGACTTCA
+C
+>YAL067W-A:115-145
+GCTTATTGTC
+TAAGCCTGAA
+TTCAGTCTGC
+T
+>chr1:1-100

src/samtools/samtools_faidx/test_data/output/test.fastq.fai

@@ -0,0 +1,2 @@
+fastq1	66	8	30	31	79
+fastq2	28	156	14	15	188

src/samtools/samtools_faidx/test_data/test.regions

@@ -0,0 +1,4 @@
+YAL069W:300-315
+YAL068C:200-230
+YAL067W-A:115-145
+chr1:1-100