Rseqc inferexperiment (#158)

* initial commit dedup

* Revert "initial commit dedup"

This reverts commit 38f586b.

* full component with two tests

* adjust arg names, container base image, test data size

---------

Co-authored-by: Robrecht Cannoodt <[email protected]>

CHANGELOG.md

@@ -17,6 +17,7 @@
 * `rsem/rsem_calculate_expression`: Calculate expression levels (PR #93).
 
 * `rseqc`:
+  - `rseqc/rseqc_inferexperiment`: Infer strandedness from sequencing reads (PR #158).
   - `rseqc/bam_stat`: Generate statistics from a bam file (PR #155).
 
 * `nanoplot`: Plotting tool for long read sequencing data and alignments (PR #95).
@@ -27,7 +28,6 @@
 
 * `cutadapt`: Fix the the non-functional `action` parameter (PR #161).
 
-
 ## MINOR CHANGES
 
 * `agat_convert_bed2gff`: change type of argument `inflate_off` from `boolean_false` to `boolean_true` (PR #160).

src/rseqc/rseqc_inferexperiment/config.vsh.yaml

@@ -0,0 +1,76 @@
+name: "rseqc_inferexperiment"
+namespace: "rseqc"
+description: |
+  Infer strandedness from sequencing reads
+links:
+  homepage: https://rseqc.sourceforge.net/
+  documentation: https://rseqc.sourceforge.net/#infer-experiment-py
+  issue_tracker: https://github.com/MonashBioinformaticsPlatform/RSeQC/issues
+  repository: https://github.com/MonashBioinformaticsPlatform/RSeQC
+references:
+  doi: 10.1093/bioinformatics/bts356
+license: GPL-3.0
+authors:
+  - __merge__: /src/_authors/emma_rousseau.yaml
+    roles: [ author, maintainer ]
+
+argument_groups:
+- name: "Input"
+  arguments: 
+  - name: "--input_file"
+    alternatives: ["-i"]
+    type: file 
+    required: true
+    description: input alignment file in BAM or SAM format
+  - name: "--refgene"
+    alternatives: ["-r"]
+    type: file 
+    required: true
+    description: Reference gene model in bed format
+
+- name: "Output"
+  arguments: 
+  - name: "--output"
+    type: file
+    direction: output
+    required: true
+    description: Output file (txt) of strandness report.
+    example: $id.strandedness.txt
+
+- name: "Options"
+  arguments:
+    - name: "--sample_size"
+      alternatives: ["-s"]
+      type: integer
+      description: |
+        Number of reads sampled from SAM/BAM file. Default: 200000
+      example: 200000
+    - name: "--mapq"
+      alternatives: ["-q"]
+      type: integer
+      description: |
+        Minimum mapping quality (phred scaled) to determine uniquely mapped reads. Default: 30
+      example: 30
+
+resources:
+  - type: bash_script
+    path: script.sh
+
+test_resources:
+  - type: bash_script
+    path: test.sh
+  - path: test_data
+
+engines:
+- type: docker
+  image: python:3.10
+  setup:
+    - type: python
+      packages: [ RSeQC ]
+    - type: docker
+      run: |
+        echo "RSeQC - infer_experiment.py: $(infer_experiment.py --version | cut -d' ' -f2)" > /var/software_versions.txt
+        
+runners: 
+- type: executable
+- type: nextflow

src/rseqc/rseqc_inferexperiment/help.txt

@@ -0,0 +1,21 @@
+```
+infer_eperiment.py --help
+```
+
+Usage: infer_experiment.py [options]
+
+
+Options:
+  --version             show program's version number and exit
+  -h, --help            show this help message and exit
+  -i INPUT_FILE, --input-file=INPUT_FILE
+                        Input alignment file in SAM or BAM format
+  -r REFGENE_BED, --refgene=REFGENE_BED
+                        Reference gene model in bed fomat.
+  -s SAMPLE_SIZE, --sample-size=SAMPLE_SIZE
+                        Number of reads sampled from SAM/BAM file.
+                        default=200000
+  -q MAP_QUAL, --mapq=MAP_QUAL
+                        Minimum mapping quality (phred scaled) for an
+                        alignment to be considered as "uniquely mapped".
+                        default=30

src/rseqc/rseqc_inferexperiment/script.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+set -eo pipefail 
+
+infer_experiment.py \
+    -i $par_input_file \
+    -r $par_refgene \
+    ${par_sample_size:+-s "${par_sample_size}"} \
+    ${par_mapq:+-q "${par_mapq}"} \
+> $par_output

src/rseqc/rseqc_inferexperiment/test.sh

@@ -0,0 +1,72 @@
+#!/bin/bash
+
+# define input and output for script
+input_bam="$meta_resources_dir/test_data/sample.bam"
+input_bed="$meta_resources_dir/test_data/test.bed12"
+output="strandedness.txt"
+
+echo ">>> Prepare test output data"
+
+cat > "$meta_resources_dir/test_data/strandedness.txt" <<EOF
+
+
+This is PairEnd Data
+Fraction of reads failed to determine: 0.0000
+Fraction of reads explained by "1++,1--,2+-,2-+": 1.0000
+Fraction of reads explained by "1+-,1-+,2++,2--": 0.0000
+EOF
+
+cat > "$meta_resources_dir/test_data/strandedness2.txt" <<EOF
+Unknown Data type
+EOF
+
+################################################################################
+# run executable and tests
+
+echo ">>> Test 1: Test with default parameters"
+
+"$meta_executable" \
+    --input_file "$input_bam" \
+    --refgene "$input_bed" \
+    --output "$output"
+
+exit_code=$?
+[[ $exit_code != 0 ]] && echo "Non zero exit code: $exit_code" && exit 1
+
+echo ">> Checking whether output can be found and has content"
+
+[ ! -f "$output" ] && echo "$output is missing" && exit 1
+[ ! -s "$output" ] && echo "$output is empty" && exit 1
+
+
+echo ">> Checking whether output is correct"
+diff "$output" "$meta_resources_dir/test_data/strandedness.txt" || { echo "Output is not correct"; exit 1; }
+
+rm "$output"
+
+################################################################################
+
+echo ">>> Test 2: Test with non-default sample size and map quality"
+
+"$meta_executable" \
+    --input_file "$input_bam" \
+    --refgene "$input_bed" \
+    --output "$output" \
+    --sample_size 150000 \
+    --mapq 90
+
+exit_code=$?
+[[ $exit_code != 0 ]] && echo "Non zero exit code: $exit_code" && exit 1
+
+echo ">> Checking whether output can be found and has content"
+
+[ ! -f "$output" ] && echo "$output is missing" && exit 1
+[ ! -s "$output" ] && echo "$output is empty" && exit 1
+
+echo ">> Checking whether output is correct"
+diff "$output" "$meta_resources_dir/test_data/strandedness2.txt" || { echo "Output is not correct"; exit 1; }
+
+
+echo "All tests passed"
+
+exit 0

src/rseqc/rseqc_inferexperiment/test_data/test.bed12

@@ -0,0 +1,4 @@
+MT192765.1	1242	1264	nCoV-2019_5_LEFT	1	+	1242	1264	0	2	10,12,	0,10,
+MT192765.1	1573	1595	nCoV-2019_6_LEFT	2	+	1573	1595	0	2	7,15,	0,7,
+MT192765.1	1623	1651	nCoV-2019_5_RIGHT	1	-	1623	1651	0	2	14,14,	0,14,
+MT192765.1	1942	1964	nCoV-2019_6_RIGHT	2	-	1942	1964	0	2	11,11	0,11,