Parcourir la source

first commit

master
lizhihui il y a 6 ans
révision
2bc4515640
8 fichiers modifiés avec 168 ajouts et 0 suppressions
  1. +13
    -0
      inputs
  2. BIN
      tasks/.DS_Store
  3. +21
    -0
      tasks/fastqc.wdl
  4. +25
    -0
      tasks/hisat2.wdl
  5. +31
    -0
      tasks/samtools.wdl
  6. +26
    -0
      tasks/stringtie.wdl
  7. +23
    -0
      tasks/trimmomatic.wdl
  8. +29
    -0
      workflow.wdl

+ 13
- 0
inputs Voir le fichier

@@ -0,0 +1,13 @@
{
"{{ project_name }}.read1": "{{ read1 }}",
"{{ project_name }}.read2": "{{ read2 }}",
"{{ project_name }}.idx": "oss://pgx-reference-data/reference/hisat2/grch38_snp_tran/",
"{{ project_name }}.gtf": "oss://pgx-reference-data/reference/annotation/Homo_sapiens.GRCh38.93.gtf",
"{{ project_name }}.idx_prefix": "genome_snp_tran",
"{{ project_name }}.hisat2.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/hisat2:v2.0.5-1-deb-cv1",
"{{ project_name }}.hisat2.cluster": "OnDemand bcs.a2.3xlarge img-ubuntu-vpc",
"{{ project_name }}.samtools.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/samtools:v1.3.1",
"{{ project_name }}.samtools.cluster": "OnDemand bcs.a2.large img-ubuntu-vpc",
"{{ project_name }}.stringtie.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/stringtie:v1.3.4",
"{{ project_name }}.stringtie.cluster": "OnDemand bcs.a2.large img-ubuntu-vpc"
}

BIN
tasks/.DS_Store Voir le fichier


+ 21
- 0
tasks/fastqc.wdl Voir le fichier

@@ -0,0 +1,21 @@
task qc {
File read
String docker
String cluster
String out_dir = "./"

command {
fastqc -o ${out_dir} ${read}
}

runtime {
docker: docker
cluster: cluster
}

# 输入文件后缀必须为.fq.gz或者.fastq.gz
output {
File html = sub(basename(read), "\\.(fastq|fq)\\.gz$", "_fastqc.html")
File zip = sub(basename(read), "\\.(fastq|fq)\\.gz$", "_fastqc.zip")
}
}

+ 25
- 0
tasks/hisat2.wdl Voir le fichier

@@ -0,0 +1,25 @@
task hisat2 {
File idx
File read_1P
File read_2P
String idx_prefix
String base = sub(basename(read_1P),"\\.\\S+$", "")
String docker
String cluster

command {
nt=$(nproc)
hisat2 -t -p $nt -x ${idx}/${idx_prefix} -1 ${read_1P} -2 ${read_2P} -S ${base}.sam
}
runtime {
docker: docker
cluster: cluster
systemDisk: "cloud_ssd 40"
dataDisk: "cloud_ssd 200 /cromwell_root/"
}

output {
File sam = base + ".sam"
}
}

+ 31
- 0
tasks/samtools.wdl Voir le fichier

@@ -0,0 +1,31 @@
task samtools {
File sam
String base = basename(sam, ".sam")
String bam = base + ".bam"
String sorted_bam = base + ".sorted.bam"
String sorted_bam_index = base + ".sorted.bam.bai"
String docker
String cluster

command <<<
set -o pipefail
set -e
/opt/conda/bin/samtools view -bS ${sam} > ${bam}
/opt/conda/bin/samtools sort -m 5000000000 ${bam} -o ${sorted_bam}
/opt/conda/bin/samtools index ${sorted_bam}
>>>

runtime {
docker: docker
cluster: cluster
systemDisk: "cloud_ssd 40"
dataDisk: "cloud_ssd 200 /cromwell_root/"
}

output {
File out_bam = sorted_bam
File out_bam_index = sorted_bam_index
}

}


+ 26
- 0
tasks/stringtie.wdl Voir le fichier

@@ -0,0 +1,26 @@
task stringtie {
File bam
File gtf
String docker
String base = basename(bam, ".sorted.bam")
String cluster

command <<<
nt=$(nproc)
mkdir ballgown
/opt/conda/bin/stringtie -e -B -p $nt -G ${gtf} -o ballgown/${base}/${base}.gtf -C ${base}.cov.ref.gtf -A ${base}.gene.abundance.txt ${bam}
>>>
runtime {
docker: docker
cluster: cluster
systemDisk: "cloud_ssd 40"
dataDisk: "cloud_ssd 150 /cromwell_root/"
}
output {
File covered_transcripts = "${base}.cov.ref.gtf"
File gene_abundance = "${base}.gene.abundance.txt"
Array[File] ballgown = ["ballgown/${base}/${base}.gtf", "ballgown/${base}/e2t.ctab", "ballgown/${base}/e_data.ctab", "ballgown/${base}/i2t.ctab", "ballgown/${base}/i_data.ctab", "ballgown/${base}/t_data.ctab"]
}
}

+ 23
- 0
tasks/trimmomatic.wdl Voir le fichier

@@ -0,0 +1,23 @@
task trimmomatic {
File read1
File read2
File adapter
String baseout
String baseout_gz = baseout + ".fq.gz"
String docker
String cluster

command {
/opt/conda/bin/trimmomatic PE -threads 20 -phred33 ${read1} ${read2} -baseout ${baseout_gz} ILLUMINACLIP:${adapter}:2:30:10:1:true HEADCROP:10 LEADING:10 TRAILING:10 SLIDINGWINDOW:4:15 MINLEN:36
}

runtime {
docker: docker
cluster: cluster
}

output {
File read_1p = baseout + "_1P.fq.gz"
File read_2p = baseout + "_2P.fq.gz"
}
}

+ 29
- 0
workflow.wdl Voir le fichier

@@ -0,0 +1,29 @@
import "./tasks/hisat2.wdl" as hisat2
import "./tasks/samtools.wdl" as samtools
import "./tasks/stringtie.wdl" as stringtie

workflow {{ project_name }} {

File read1
File read2
File idx
String idx_prefix
File gtf
call hisat2.hisat2 as hisat2 {
input: idx=idx, idx_prefix=idx_prefix, read_1P=read1, read_2P=read2
}

call samtools.samtools as samtools {
input: sam = hisat2.sam
}

call stringtie.stringtie as stringtie {
input: gtf = gtf, bam = samtools.out_bam
}
}



Chargement…
Annuler
Enregistrer