Merge branch 'master' into r-ampir

modules/cellranger/Dockerfile

@@ -1,4 +1,4 @@
-# Dockerfile to create container with Cell Ranger v6.1.2
+# Dockerfile to create container with Cell Ranger v7.0.0
 # Push to nfcore/cellranger:<VER>
 
 FROM continuumio/miniconda3:4.8.2

modules/cellranger/count/main.nf

@@ -5,7 +5,7 @@ process CELLRANGER_COUNT {
     if (params.enable_conda) {
         exit 1, "Conda environments cannot be used when using the Cell Ranger tool. Please use docker or singularity containers."
     }
-    container "nfcore/cellranger:6.1.2"
+    container "nfcore/cellranger:7.0.0"
 
     input:
     tuple val(meta), path(reads)

modules/cellranger/mkfastq/Dockerfile

@@ -1,4 +1,4 @@
-# Dockerfile to create container with Cell Ranger v6.1.2 and bcl2fastq v2.20.0
+# Dockerfile to create container with Cell Ranger v7.0.0 and bcl2fastq v2.20.0
 # Push to nfcore/cellrangermkfastq:<VER>
 
 FROM continuumio/miniconda3:4.8.2
@@ -17,7 +17,7 @@ RUN apt-get update --allow-releaseinfo-change \
 
 # Copy pre-downloaded bcl2fastq2 and cellranger file
 ENV BCL2FASTQ2_VER=v2-20-0-linux-x86-64 \
-    CELLRANGER_VER=6.1.2
+    CELLRANGER_VER=7.0.0
 COPY bcl2fastq2-$BCL2FASTQ2_VER.zip /tmp/bcl2fastq2-$BCL2FASTQ2_VER.zip
 COPY cellranger-$CELLRANGER_VER.tar.gz /opt/cellranger-$CELLRANGER_VER.tar.gz
 

modules/cellranger/mkfastq/main.nf

@@ -5,7 +5,7 @@ process CELLRANGER_MKFASTQ {
     if (params.enable_conda) {
         exit 1, "Conda environments cannot be used when using the Cell Ranger tool. Please use docker or singularity containers."
     }
-    container "nfcore/cellrangermkfastq:6.1.2"
+    container "nfcore/cellrangermkfastq:7.0.0"
 
     input:
     path bcl

modules/cellranger/mkgtf/main.nf

@@ -5,7 +5,7 @@ process CELLRANGER_MKGTF {
     if (params.enable_conda) {
         exit 1, "Conda environments cannot be used when using the Cell Ranger tool. Please use docker or singularity containers."
     }
-    container "nfcore/cellranger:6.1.2"
+    container "nfcore/cellranger:7.0.0"
 
     input:
     path gtf

modules/cellranger/mkref/main.nf

@@ -5,7 +5,7 @@ process CELLRANGER_MKREF {
     if (params.enable_conda) {
         exit 1, "Conda environments cannot be used when using the Cell Ranger tool. Please use docker or singularity containers."
     }
-    container "nfcore/cellranger:6.1.2"
+    container "nfcore/cellranger:7.0.0"
 
     input:
     path fasta

modules/sexdeterrmine/main.nf

@@ -0,0 +1,40 @@
+process SEXDETERRMINE {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::sexdeterrmine=1.1.2" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/sexdeterrmine:1.1.2--hdfd78af_1':
+        'quay.io/biocontainers/sexdeterrmine:1.1.2--hdfd78af_1' }"
+
+    input:
+    tuple val(meta), path(depth)
+    path sample_list_file
+
+    output:
+    tuple val(meta), path("*.json"), emit: json
+    tuple val(meta), path("*.tsv") , emit: tsv
+    path "versions.yml"            , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def sample_list = sample_list_file ? '-f ${sample_list_file}' : ''
+    if ("$depth" == "${prefix}.tsv") error "Input depth and output TSV names are the same, set prefix in module configuration to disambiguate!"
+
+    """
+    sexdeterrmine \\
+        -I $depth \\
+        $sample_list \\
+        $args \\
+        > ${prefix}.tsv
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        sexdeterrmine: \$(echo \$(sexdeterrmine --version 2>&1))
+    END_VERSIONS
+    """
+}

modules/sexdeterrmine/meta.yml

@@ -0,0 +1,48 @@
+name: "sexdeterrmine"
+description: Calculate the relative coverage on the Gonosomes vs Autosomes from the output of samtools depth, with error bars.
+keywords:
+  - sex determination
+  - genetic sex
+  - relative coverage
+  - ancient dna
+tools:
+  - "sexdeterrmine":
+      description: "A python script carry out calculate the relative coverage of X and Y chromosomes, and their associated error bars, out of capture data."
+      homepage: "https://github.com/TCLamnidis/Sex.DetERRmine"
+      documentation: "https://github.com/TCLamnidis/Sex.DetERRmine/README.md"
+      tool_dev_url: "https://github.com/TCLamnidis/Sex.DetERRmine"
+      doi: "https://doi.org/10.1038/s41467-018-07483-5"
+      licence: "['GPL v3']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - depth:
+      type: file
+      description: Output from samtools depth (with header)
+      pattern: "*"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - json:
+      type: file
+      description: JSON formatted table of relative coverages on the X and Y, with associated error bars.
+      pattern: "*.json"
+  - tsv:
+      type: file
+      description: TSV table of relative coverages on the X and Y, with associated error bars.
+      pattern: "*.tsv"
+
+authors:
+  - "@TCLamnidis"

tests/config/pytest_modules.yml

@@ -1855,6 +1855,10 @@ seqwish/induce:
   - modules/seqwish/induce/**
   - tests/modules/seqwish/induce/**
 
+sexdeterrmine:
+  - modules/sexdeterrmine/**
+  - tests/modules/sexdeterrmine/**
+
 shasum:
   - modules/shasum/**
   - tests/modules/shasum/**

tests/config/test_data.config

@@ -232,10 +232,11 @@ params {
                 test2_paired_end_umi_unsorted_bam              = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/umi/test2.paired_end.umi_unsorted.bam"
                 test2_paired_end_umi_unsorted_tagged_bam       = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/umi/test2.paired_end.unsorted_tagged.bam"
 
-
                 mitochon_standin_recalibrated_sorted_bam       = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/mitochon_standin.recalibrated.sorted.bam"
                 mitochon_standin_recalibrated_sorted_bam_bai   = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/mitochon_standin.recalibrated.sorted.bam.bai"
 
+                test3_single_end_markduplicates_sorted_bam       = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test3.single_end.markduplicates.sorted.bam"
+
                 test_paired_end_sorted_cram                      = "${test_data_dir}/genomics/homo_sapiens/illumina/cram/test.paired_end.sorted.cram"
                 test_paired_end_sorted_cram_crai                 = "${test_data_dir}/genomics/homo_sapiens/illumina/cram/test.paired_end.sorted.cram.crai"
                 test_paired_end_markduplicates_sorted_cram       = "${test_data_dir}/genomics/homo_sapiens/illumina/cram/test.paired_end.markduplicates.sorted.cram"

tests/modules/cellranger/count/test.yml

@@ -1,19 +1,65 @@
 - name: cellranger count test_cellranger_count
-  command: nextflow run tests/modules/cellranger/count -entry test_cellranger_count -c tests/config/nextflow.config -c tests/modules/cellranger/count/nextflow.config
+  command: nextflow run ./tests/modules/cellranger/count -entry test_cellranger_count -c ./tests/config/nextflow.config  -c ./tests/modules/cellranger/count/nextflow.config
   tags:
-    - cellranger
     - cellranger/count
+    - cellranger
   files:
+    - path: output/cellranger/genome.filtered.gtf
+      md5sum: a8b8a7b5039e05d3a9cf9151ea138b5b
+    - path: output/cellranger/homo_sapiens_chr22_reference/fasta/genome.fa
+      md5sum: f315020d899597c1b57e5fe9f60f4c3e
+    - path: output/cellranger/homo_sapiens_chr22_reference/fasta/genome.fa.fai
+      md5sum: 3520cd30e1b100e55f578db9c855f685
+    - path: output/cellranger/homo_sapiens_chr22_reference/genes/genes.gtf.gz
+      md5sum: d1e05cd46684fa26d852b6bc9f05e31f
+    - path: output/cellranger/homo_sapiens_chr22_reference/reference.json
+      md5sum: 8405fd7f527a944eafb9c2909045840b
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/Genome
+      md5sum: 897cec2d191945335f8b320438bd9135
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/SA
+      md5sum: 7961129ac5d0e1706105be1d31c6b30c
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/SAindex
+      md5sum: dcceb480b30cda93fb8c63ddc339093b
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/chrLength.txt
+      md5sum: c81f40f27e72606d7d07097c1d56a5b5
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/chrName.txt
+      md5sum: 5ae68a67b70976ee95342a7451cb5af1
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/chrNameLength.txt
+      md5sum: b190587cae0531f3cf25552d8aa674db
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/chrStart.txt
+      md5sum: bc73df776dd3d5bb9cfcbcba60880519
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/exonGeTrInfo.tab
+      md5sum: 9129691eeb4ed0d02b17be879fa3edb0
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/exonInfo.tab
+      md5sum: 209b82f0683efd03e17d2c729676554f
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/geneInfo.tab
+      md5sum: 02a8f4575bdfcd4a42b4d8d07f2e9369
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/genomeParameters.txt
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/sjdbInfo.txt
+      md5sum: 1082ab459363b3f2f7aabcef0979c1ed
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/sjdbList.fromGTF.out.tab
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/sjdbList.out.tab
+    - path: output/cellranger/homo_sapiens_chr22_reference/star/transcriptInfo.tab
+      md5sum: cedcb5f4e7d97bc548cd5daa022e092c
     - path: output/cellranger/sample-123/outs/filtered_feature_bc_matrix.h5
+      md5sum: f8b6b7cc8248151a98c46d4ebec450c6
+    - path: output/cellranger/sample-123/outs/filtered_feature_bc_matrix/barcodes.tsv.gz
+    - path: output/cellranger/sample-123/outs/filtered_feature_bc_matrix/features.tsv.gz
+    - path: output/cellranger/sample-123/outs/filtered_feature_bc_matrix/matrix.mtx.gz
     - path: output/cellranger/sample-123/outs/metrics_summary.csv
       md5sum: 707df0f101d479d93f412ca74f9c4131
     - path: output/cellranger/sample-123/outs/molecule_info.h5
-      md5sum: 0e56836ef0725f2ab05f56ca5a71e55b
+      md5sum: a13bd7425f441c8d0eac8ffc50082996
     - path: output/cellranger/sample-123/outs/possorted_genome_bam.bam
       md5sum: 15441da9cfceea0bb48c8b66b1b860df
     - path: output/cellranger/sample-123/outs/possorted_genome_bam.bam.bai
       md5sum: 7c3d49c77016a09535aff61a027f750c
-    - path: output/cellranger/sample-123/outs/raw_feature_bc_matrix
     - path: output/cellranger/sample-123/outs/raw_feature_bc_matrix.h5
-      md5sum: cdad1cd7b215d7137cf92515e81a8525
+      md5sum: a5290f3e300a4070f3d68a0c2e215f54
+    - path: output/cellranger/sample-123/outs/raw_feature_bc_matrix/barcodes.tsv.gz
+      md5sum: 5cc39ef0c7ac85f2b758b164aabf9157
+    - path: output/cellranger/sample-123/outs/raw_feature_bc_matrix/features.tsv.gz
+      md5sum: 07d497c7ce3e22f374af7b2cf9b97d72
+    - path: output/cellranger/sample-123/outs/raw_feature_bc_matrix/matrix.mtx.gz
+      md5sum: bdce94a51f16e22d40301724080b76ee
     - path: output/cellranger/sample-123/outs/web_summary.html

tests/modules/cellranger/mkfastq/test.yml

@@ -5,7 +5,6 @@
     - cellranger/mkfastq
   files:
     - path: output/cellranger/cellranger-tiny-bcl-1/outs/fastq_path/fake_file.fastq.gz
-      md5sum: d41d8cd98f00b204e9800998ecf8427e
 - name: cellranger mkfastq test_cellranger_mkfastq_illumina
   command: nextflow run tests/modules/cellranger/mkfastq -entry test_cellranger_mkfastq_illumina -c tests/config/nextflow.config -c ./tests/modules/cellranger/mkfastq/nextflow.config -stub-run
   tags:
@@ -13,4 +12,3 @@
     - cellranger/mkfastq
   files:
     - path: output/cellranger/cellranger-tiny-bcl-1/outs/fastq_path/fake_file.fastq.gz
-      md5sum: d41d8cd98f00b204e9800998ecf8427e

tests/modules/cellranger/mkgtf/test.yml

@@ -1,8 +1,8 @@
 - name: cellranger mkgtf test_cellranger_mkgtf
-  command: nextflow run tests/modules/cellranger/mkgtf -entry test_cellranger_mkgtf -c tests/config/nextflow.config -c tests/modules/cellranger/mkgtf/nextflow.config
+  command: nextflow run ./tests/modules/cellranger/mkgtf -entry test_cellranger_mkgtf -c ./tests/config/nextflow.config  -c ./tests/modules/cellranger/mkgtf/nextflow.config
   tags:
-    - cellranger
     - cellranger/mkgtf
+    - cellranger
   files:
     - path: output/cellranger/genome.filtered.gtf
       md5sum: a8b8a7b5039e05d3a9cf9151ea138b5b

tests/modules/cellranger/mkref/test.yml

@@ -1,8 +1,8 @@
 - name: cellranger mkref test_cellranger_mkref
-  command: nextflow run ./tests/modules/cellranger/mkref -entry test_cellranger_mkref -c ./tests/config/nextflow.config -c ./tests/modules/cellranger/mkref/nextflow.config
+  command: nextflow run ./tests/modules/cellranger/mkref -entry test_cellranger_mkref -c ./tests/config/nextflow.config  -c ./tests/modules/cellranger/mkref/nextflow.config
   tags:
-    - cellranger
     - cellranger/mkref
+    - cellranger
   files:
     - path: output/cellranger/homo_sapiens_chr22_reference/fasta/genome.fa
       md5sum: f315020d899597c1b57e5fe9f60f4c3e
@@ -11,7 +11,7 @@
     - path: output/cellranger/homo_sapiens_chr22_reference/genes/genes.gtf.gz
       md5sum: 6d9b5f409bfea95022bc25b9590e194e
     - path: output/cellranger/homo_sapiens_chr22_reference/reference.json
-      md5sum: 5d8d1669cd251433505f183e1c9ed6bc
+      md5sum: 6cc817f0923062e780e6573806840cea
     - path: output/cellranger/homo_sapiens_chr22_reference/star/Genome
       md5sum: 22102926fadf5890e905ca71b2da3f35
     - path: output/cellranger/homo_sapiens_chr22_reference/star/SA

tests/modules/sexdeterrmine/main.nf

@@ -0,0 +1,16 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SAMTOOLS_DEPTH } from '../../../modules/samtools/depth/main.nf'
+include { SEXDETERRMINE } from '../../../modules/sexdeterrmine/main.nf'
+
+workflow test_sexdeterrmine {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test3_single_end_markduplicates_sorted_bam'], checkIfExists: true) ]
+
+    SAMTOOLS_DEPTH ( input )
+    SEXDETERRMINE ( SAMTOOLS_DEPTH.out.tsv, [] )
+}

tests/modules/sexdeterrmine/nextflow.config

@@ -0,0 +1,12 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName:SAMTOOLS_DEPTH {
+        ext.args = "-H"
+    }
+
+    withName:SEXDETERRMINE {
+        ext.prefix = { "${meta.id}_sexdet" }
+    }
+}

tests/modules/sexdeterrmine/test.yml

@@ -0,0 +1,15 @@
+- name: sexdeterrmine test_sexdeterrmine
+  command: nextflow run tests/modules/sexdeterrmine -entry test_sexdeterrmine -c tests/config/nextflow.config
+  tags:
+    - sexdeterrmine
+  files:
+    - path: output/samtools/test.tsv
+      md5sum: fa2992ca1ea93a6e1b3e838476191935
+    - path: output/samtools/versions.yml
+      md5sum: dbd04b700335c8ad236bd667254c8dd8
+    - path: output/sexdeterrmine/sexdeterrmine.json
+      md5sum: bafb2419bb8630eda29a251c20e97166
+    - path: output/sexdeterrmine/test_sexdet.tsv
+      md5sum: 1cf8a2b97b38353eb97a96ab872dcca9
+    - path: output/sexdeterrmine/versions.yml
+      md5sum: 077361101e8e7997aec3da8a01e59eee

tests/test_versions_yml.py

@@ -16,9 +16,9 @@ def _get_workflow_names():
         # test_config = yaml.safe_load(f.read_text())
         test_config = yaml.load(f.read_text(), Loader=yaml.BaseLoader)
         for workflow in test_config:
-            # https://github.com/nf-core/modules/pull/1242 - added to cover tests 
+            # https://github.com/nf-core/modules/pull/1242 - added to cover tests
             # that expect an error and therefore will not generate a versions.yml
-            if 'exit_code' not in workflow: 
+            if 'exit_code' not in workflow:
                 yield workflow["name"]
 
 
@@ -56,5 +56,5 @@ def test_ensure_valid_version_yml(workflow_dir):
     assert len(software_versions), "There must be at least one version emitted."
     for tool, version in software_versions.items():
         assert re.match(
-            r"^\d+.*", str(version)
+            r"^\d.*|^[a-f0-9]{40}$", str(version)
-        ), f"Version number for {tool} must start with a number. "
+        ), f"Version number for {tool} must start with a number, or be a Git SHA commit id. "