cellranger count: start with L001 when renaming files (nf-core#4901)

* cellranger count: start with L001 when renaming files

Previously it started with L000 which was still recognized by cellranger, but there would never be a file named L000 generated by demultiplexing.

* automated pytest migration

* Update nf-test file

* Update nf-test snapshot

* [automated] Fix linting with Prettier

* Fix module lint

* Update nf-test

* exclude empty file from snap

* Fix nf-test

---------

Co-authored-by: nf-core-bot <[email protected]>
Co-authored-by: Alexander Peltzer <[email protected]>

modules/nf-core/cellranger/count/main.nf

@@ -32,7 +32,7 @@ process CELLRANGER_COUNT {
     def prefix = task.ext.prefix ?: "${meta.id}"
     """
     mkdir -p "${prefix}/outs/"
-    touch ${prefix}/outs/fake_file.txt
+    echo "$prefix" > ${prefix}/outs/fake_file.txt
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/nf-core/cellranger/count/meta.yml

@@ -10,7 +10,8 @@ tools:
       homepage: https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger
       documentation: https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/using/tutorial_ov
       tool_dev_url: https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/using/tutorial_ov
-      licence: 10x Genomics EULA
+      licence:
+        - 10x Genomics EULA
 input:
   - meta:
       type: map

modules/nf-core/cellranger/count/templates/cellranger_count.py

@@ -36,7 +36,7 @@ def chunk_iter(seq, size):
 # do not match "SRR12345", "file_INFIXR12", etc
 filename_pattern =  r'([^a-zA-Z0-9])R1([^a-zA-Z0-9])'
 
-for i, (r1, r2) in enumerate(chunk_iter(fastqs, 2)):
+for i, (r1, r2) in enumerate(chunk_iter(fastqs, 2), start=1):
     # double escapes are required because nextflow processes this python 'template'
     if re.sub(filename_pattern, r'\\1R2\\2', r1.name) != r2.name:
         raise AssertionError(

modules/nf-core/cellranger/count/tests/main.nf.test

@@ -0,0 +1,101 @@
+nextflow_process {
+
+    name "Test Process CELLRANGER_COUNT"
+    script "../main.nf"
+    process "CELLRANGER_COUNT"
+
+    tag "modules"
+    tag "modules_nfcore"
+    tag "cellranger"
+    tag "cellranger/count"
+    tag "cellranger/mkref"
+    tag "cellranger/mkgtf"
+
+    setup {
+        config "./nextflow.config"
+
+        run("CELLRANGER_MKGTF") {
+            script "../../mkgtf/main.nf"
+            process {
+                """
+                input[0] = file(params.test_data['homo_sapiens']['genome']['genome_gtf'], checkIfExists: true)
+                """
+            }
+        }
+        run("CELLRANGER_MKREF") {
+            script "../../mkref/main.nf"
+            process {
+                """
+                input[0] = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+                input[1] = CELLRANGER_MKGTF.out.gtf
+                input[2] = "homo_sapiens_chr22_reference"
+                """
+            }
+        }
+    }
+
+    test("10x example file") {
+        config "./nextflow.config"
+        when {
+            process {
+                """
+                input[0] =  [
+                    [ id:'test_10x', single_end:false, strandedness:'auto'  ], // meta map
+                    [
+                        file(params.test_data['homo_sapiens']['10xgenomics']['cellranger']['test_10x_5k_cmvpos_tcells_gex1_fastq_1_gz'], checkIfExists: true),
+                        file(params.test_data['homo_sapiens']['10xgenomics']['cellranger']['test_10x_5k_cmvpos_tcells_gex1_fastq_2_gz'], checkIfExists: true)
+                    ]
+                ]
+                input[1] = CELLRANGER_MKREF.out.reference
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                {
+                  assert snapshot(
+                    process.out.versions,
+                    process.out.outs[0][1].findAll { file(it).name !in [
+                        'web_summary.html', // unstable checksum
+                        'barcodes.tsv.gz' // empty file in test data -> would raise linting error
+                    ]}
+                  ).match()
+                },
+                { assert file(process.out.outs.get(0).get(1).find { file(it).name == 'web_summary.html' }).exists() },
+                { assert file(process.out.outs.get(0).get(1).find { file(it).name == 'barcodes.tsv.gz' }).exists() }
+            )
+        }
+
+    }
+
+    test("10x example file - stub") {
+
+        options "-stub"
+
+        when {
+            process {
+                """
+                input[0] =  [
+                    [ id:'test_10x', single_end:false, strandedness:'auto'  ], // meta map
+                    [
+                        file(params.test_data['homo_sapiens']['10xgenomics']['cellranger']['test_10x_5k_cmvpos_tcells_gex1_fastq_1_gz'], checkIfExists: true),
+                        file(params.test_data['homo_sapiens']['10xgenomics']['cellranger']['test_10x_5k_cmvpos_tcells_gex1_fastq_2_gz'], checkIfExists: true)
+                    ]
+                ]
+                input[1] = CELLRANGER_MKREF.out.reference
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+
+    }
+
+}

tests/modules/nf-core/cellranger/count/nextflow.config → modules/nf-core/cellranger/count/tests/nextflow.config

@@ -1,7 +1,4 @@
 process {
-
-    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
-
     withName: CELLRANGER_MKGTF {
         ext.args = '--attribute=gene_biotype:protein_coding \
             --attribute=gene_biotype:lincRNA \
@@ -20,12 +17,8 @@ process {
             --attribute=gene_biotype:TR_J_gene \
             --attribute=gene_biotype:TR_J_pseudogene \
             --attribute=gene_biotype:TR_C_gene'
-
-
     }
-
     withName: CELLRANGER_COUNT {
         ext.args = '--chemistry SC3Pv3'
     }
-
 }

modules/nf-core/cellranger/count/tests/tags.yml

@@ -0,0 +1,2 @@
+cellranger/count:
+  - "modules/nf-core/cellranger/count/**"

tests/config/pytest_modules.yml

@@ -301,13 +301,6 @@ canu:
 cdhit/cdhitest:
   - modules/nf-core/cdhit/cdhitest/**
   - tests/modules/nf-core/cdhit/cdhitest/**
-cellranger/count:
-  - modules/nf-core/cellranger/count/**
-  - tests/modules/nf-core/cellranger/count/**
-  - modules/nf-core/cellranger/mkref/**
-  - tests/modules/nf-core/cellranger/mkref/**
-  - modules/nf-core/cellranger/mkgtf/**
-  - tests/modules/nf-core/cellranger/mkgtf/**
 cellranger/mkfastq:
   - modules/nf-core/cellranger/mkfastq/**
   - tests/modules/nf-core/cellranger/mkfastq/**