From 943e5e521f5c8dbc0feba007ea1d7d6ff6781e88 Mon Sep 17 00:00:00 2001
From: Dominik Brilhaus <brilhaus@nfdi4plants.org>
Date: Tue, 18 Mar 2025 09:57:56 +0100
Subject: [PATCH] switch kallisto input to array of records
---
runs/kallisto/run.cwl | 19 ++++---
runs/kallisto/run.yml | 57 ++++++++++---------
workflows/kallisto/workflow.cwl | 99 +++++++++++++++++----------------
3 files changed, 92 insertions(+), 83 deletions(-)
diff --git a/runs/kallisto/run.cwl b/runs/kallisto/run.cwl
index 7ab4d83..3ec2a22 100644
--- a/runs/kallisto/run.cwl
+++ b/runs/kallisto/run.cwl
@@ -4,16 +4,20 @@ class: Workflow
requirements:
SubworkflowFeatureRequirement: {}
+ MultipleInputFeatureRequirement: {}
inputs:
IndexInput: File[]
- InputReadsMultipleSamples:
- type:
+ sampleRecord:
+ type:
type: array
- items:
- type: array
- items: File
- numUnderscoresFileName: int
+ items:
+ type: record
+ fields:
+ readsOfOneSample:
+ type: File[]
+ sampleName:
+ type: string?
isSingle: boolean
FragmentLength: double?
StandardDeviation: double?
@@ -25,8 +29,7 @@ steps:
run: ../../workflows/kallisto/workflow.cwl
in:
IndexInput: IndexInput
- InputReadsMultipleSamples: InputReadsMultipleSamples
- numUnderscoresFileName: numUnderscoresFileName
+ sampleRecord: sampleRecord
isSingle: isSingle
FragmentLength: FragmentLength
StandardDeviation: StandardDeviation
diff --git a/runs/kallisto/run.yml b/runs/kallisto/run.yml
index 1d147f9..541e170 100644
--- a/runs/kallisto/run.yml
+++ b/runs/kallisto/run.yml
@@ -5,44 +5,47 @@ IndexInput:
format: edam:format_1929 # FASTA
## Fastq files to be mapped
-# The `InputReadsMultipleSamples` looks more complicated than needed
-# It's an array of File arrays to generically allow multiple fastq files per sample
+# The `readsOfOneSample` looks more complicated than needed
+# It's an array of records (each with one or mupltiple files and a sample name) to generically allow multiple fastq files per sample
-InputReadsMultipleSamples:
- -
+sampleRecord:
+ - readsOfOneSample:
- class: File
path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_097_CAMMD_CAGATC_L001_R1_001.fastq.gz
- format: edam:format_1930 # FASTQ
- -
+ format: edam:format_1930
+ sampleName: 'DB_097'
+ - readsOfOneSample:
+ - class: File
+ path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_163_reC3MD_GTGAAA_L001_R1_001.fastq.gz
+ format: edam:format_1930
+ sampleName: 'DB_163'
+ - readsOfOneSample:
- class: File
path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_099_CAMMD_CTTGTA_L001_R1_001.fastq.gz
- format: edam:format_1930 # FASTQ
- # -
- # - class: File
- # path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_103_CAMMD_AGTCAA_L001_R1_001.fastq.gz
- # format: edam:format_1930 # FASTQ
- # -
- # - class: File
- # path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_161_reC3MD_GTCCGC_L001_R1_001.fastq.gz
- # format: edam:format_1930 # FASTQ
- # -
- # - class: File
- # path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_163_reC3MD_GTGAAA_L001_R1_001.fastq.gz
- # format: edam:format_1930 # FASTQ
- # -
- # - class: File
- # path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_165_re-C3MD_GTGAAA_L002_R1_001.fastq.gz
- # format: edam:format_1930 # FASTQ
-
-numUnderscoresFileName: 2
+ format: edam:format_1930
+ sampleName: 'DB_099'
+ - readsOfOneSample:
+ - class: File
+ path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_103_CAMMD_AGTCAA_L001_R1_001.fastq.gz
+ format: edam:format_1930
+ sampleName: 'DB_103'
+ - readsOfOneSample:
+ - class: File
+ path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_161_reC3MD_GTCCGC_L001_R1_001.fastq.gz
+ format: edam:format_1930
+ sampleName: 'DB_161'
+ - readsOfOneSample:
+ - class: File
+ path: ../../assays/Talinum_RNASeq_minimal/dataset/DB_165_re-C3MD_GTGAAA_L002_R1_001.fastq.gz
+ format: edam:format_1930
+ sampleName: 'DB_165'
### Kallisto quant Parameters
isSingle: true
FragmentLength: 200
StandardDeviation: 20
BootstrapSamples: 30
-
-resultsFolder: test
+resultsFolder: kallisto_results
$namespaces:
edam: https://edamontology.org/
\ No newline at end of file
diff --git a/workflows/kallisto/workflow.cwl b/workflows/kallisto/workflow.cwl
index ea6b79d..4ce5617 100644
--- a/workflows/kallisto/workflow.cwl
+++ b/workflows/kallisto/workflow.cwl
@@ -1,3 +1,4 @@
+#!/usr/bin/env cwl-runner
cwlVersion: v1.2
class: Workflow
@@ -7,56 +8,58 @@ requirements:
InlineJavascriptRequirement: {}
inputs:
- IndexInput: File[]
- sampleRecord:
- type:
- type: array
- items:
- type: record
- fields:
- readsOfOneSample:
- type: File[]
- sampleName:
- type: string?
- isSingle: boolean
- FragmentLength: double?
- StandardDeviation: double?
- BootstrapSamples: int?
- resultsFolder: string
+ IndexInput: File[]
+ sampleRecord:
+ type:
+ type: array
+ items:
+ type: record
+ fields:
+ readsOfOneSample:
+ type: File[]
+ sampleName:
+ type: string?
+ isSingle: boolean
+ FragmentLength: double?
+ StandardDeviation: double?
+ BootstrapSamples: int?
+ resultsFolder: string
steps:
- index:
- run: kallisto-index.cwl
- in:
- InputFiles: IndexInput
- IndexName:
- source: IndexInput
- valueFrom: $(self[0].nameroot)
- out: [index]
- quant:
- run: kallisto-quant.cwl
- scatter: [InputReads, QuantOutfolder]
- scatterMethod: dotproduct
- in:
- InputReads:
- source: sampleRecord
- valueFrom: $(self.readsOfOneSample)
- QuantOutfolder:
- source: sampleRecord
- valueFrom: $(self.sampleName)
- Index: index/index
- isSingle: isSingle
- FragmentLength: FragmentLength
- StandardDeviation: StandardDeviation
- BootstrapSamples: BootstrapSamples
- out: [outFolder]
- collectResults:
- run: ./yield-dirInDestination.cwl
- scatter: inDir
- in:
- inDir: quant/outFolder
- destinationDir: resultsFolder
- out: [outDir]
+ index:
+ run: kallisto-index.cwl
+ in:
+ InputFiles: IndexInput
+ IndexName:
+ source: IndexInput
+ valueFrom: $(self[0].nameroot)
+ out: [index]
+
+ quant:
+ run: kallisto-quant.cwl
+ scatter: [InputReads, QuantOutfolder]
+ scatterMethod: dotproduct
+ in:
+ InputReads:
+ source: sampleRecord
+ valueFrom: $(self.readsOfOneSample)
+ QuantOutfolder:
+ source: sampleRecord
+ valueFrom: $(self.sampleName)
+ Index: index/index
+ isSingle: isSingle
+ FragmentLength: FragmentLength
+ StandardDeviation: StandardDeviation
+ BootstrapSamples: BootstrapSamples
+ out: [outFolder]
+
+ collectResults:
+ run: ./yield-dirInDestination.cwl
+ scatter: inDir
+ in:
+ inDir: quant/outFolder
+ destinationDir: resultsFolder
+ out: [outDir]
outputs:
finalOut:
--
GitLab