Merge pull request #19 from yyoshiaki/developmentv1.2

Developmentv1.2

README.md

@@ -1,4 +1,4 @@
-# VIRTUS : VIRal Transcript Usage Sensor v1.1 <img src="https://github.com/yyoshiaki/VIRTUS/raw/master/img/VIRTUS.jpg" width="20%" align="right" />
+# VIRTUS : VIRal Transcript Usage Sensor v1.2 <img src="https://github.com/yyoshiaki/VIRTUS/raw/master/img/VIRTUS.jpg" width="20%" align="right" />
 
 Virus transcript detection and quantification using normal human RNAseq. VIRTUS is the first tool to detect viral transcripts considering their splicing event rather than the viral genome copy number. VIRTUS can be applied to both bulk RNAseq and single-cell RNAseq. The virus reference covers 762 viruses including SARS-CoV-2 (cause of COVID-19). The workflow is implemented by [Common Workflow Language](https://www.commonwl.org/) and [Rabix](https://rabix.io/). You can specify each parameter individually or give `yaml` or `json` file which describes all the parameter information. In detail, check [the CWL User Guide](http://www.commonwl.org/user_guide/) out. 
 
@@ -16,7 +16,7 @@ Yoshiaki Yasumizu ([yyasumizu@ifrec.osaka-u.ac.jp](yyasumizu@ifrec.osaka-u.ac.jp
 ## Citation
 
 VIRTUS: a pipeline for comprehensive virus analysis from conventional RNA-seq data
-Yoshiaki Yasumizu, Atsushi Hara, Shimon Sakaguchi, Naganari Ohkura. *bioRxiv* 2020.05.08.085308; doi: https://doi.org/10.1101/2020.05.08.085308
+Yasumizu, Yoshiaki, Atsushi Hara, Shimon Sakaguchi, and Naganari Ohkura. 2020. “OUP Accepted Manuscript.” Edited by Jan Gorodkin. *Bioinformatics*, October. https://doi.org/10.1093/bioinformatics/btaa859.
 
 ## Acknowledgement
 

test/test.sh

@@ -45,4 +45,6 @@ if [[ ! -e ./SRR8315715_1.fastq.gz ]]; then
 fi
 cwltool --rm-tmpdir ../../workflow/VIRTUS.SE.cwl ../../workflow/VIRTUS.SE.job.yaml
 cwltool --rm-tmpdir ../../workflow/VIRTUS.SE.singlevirus.cwl ../../workflow/VIRTUS.SE.singlevirus.job.yaml
-cd ..
+cd ..
+
+echo Successfully completed test.sh!

tool/fastq_pair/fastq_pair.cwl

@@ -0,0 +1,36 @@
+class: CommandLineTool
+cwlVersion: v1.0
+$namespaces:
+  sbg: 'https://www.sevenbridges.com/'
+id: fastq_pair
+baseCommand:
+  - fastq_pair
+inputs:
+  - id: fq1
+    type: File
+    inputBinding:
+      position: 0
+  - id: fq2
+    type: File
+    inputBinding:
+      position: 0
+outputs:
+  - id: fq1_paired
+    type: File
+    outputBinding:
+      glob: $(inputs.fq1.basename).paired.fq
+  - id: fq2_paired
+    type: File
+    outputBinding:
+      glob: $(inputs.fq2.basename).paired.fq
+label: fastq_pair
+requirements:
+  - class: DockerRequirement
+    dockerPull: 'quay.io/biocontainers/fastq-pair:1.0--he1b5a44_1'
+  - class: InlineJavascriptRequirement
+  - class: InitialWorkDirRequirement
+    listing:
+      - entry: $(inputs.fq1)
+        writable: true
+      - entry: $(inputs.fq2)
+        writable: true

tool/fastq_pair/fastq_pair.job.yaml

@@ -0,0 +1,6 @@
+fq1:
+  class: File
+  path: /Users/yyasumizu/bioinformatics/tmp/kz_1.fq
+fq2:
+  class: File
+  path: /Users/yyasumizu/bioinformatics/tmp/kz_2.fq

tool/kz_filter/Dockerfile

@@ -0,0 +1,4 @@
+FROM conda/miniconda3
+USER root
+
+RUN conda install -y -c eclarke komplexity

tool/kz_filter/kz_filter.cwl

@@ -0,0 +1,37 @@
+class: CommandLineTool
+cwlVersion: v1.0
+$namespaces:
+  sbg: 'https://www.sevenbridges.com/'
+id: kz_filter
+baseCommand:
+  - kz
+inputs:
+  - 'sbg:toolDefaultValue': '0.1'
+    id: threshold
+    type: float?
+    inputBinding:
+      prefix: '--threshold'
+      shellQuote: false
+      position: 0
+  - id: input_fq
+    type: File
+  - id: output_fq
+    type: string
+outputs:
+  - id: output
+    type: File?
+    outputBinding:
+      glob: $(inputs.output_fq)
+label: kz-filter
+arguments:
+  - prefix: ''
+    shellQuote: false
+    position: 0
+    valueFrom: '--filter'
+requirements:
+  - class: ShellCommandRequirement
+  - class: DockerRequirement
+    dockerPull: 'yyasumizu/ko:0.1'
+  - class: InlineJavascriptRequirement
+stdin: $(inputs.input_fq.path)
+stdout: $(inputs.output_fq)

tool/kz_filter/kz_filter.job.yaml

@@ -0,0 +1,6 @@
+input_fq:
+  class: File
+  path: >-
+    /home/yyasumizu/media32TB/bioinformatics/practice/VIRTUS/test/ERR3240275/unmapped_1.fq
+output_fq: unmapped_kz0.1.fq
+threshold: 0.1

tool/samtools/bam_filter_polyx.cwl

@@ -20,5 +20,5 @@ label: bam_filter_polyX
 requirements:
   - class: ShellCommandRequirement
   - class: DockerRequirement
-    dockerPull: 'yyasumizu/bam_filter_polyx:1.1'
+    dockerPull: 'yyasumizu/bam_filter_polyx:1.2'
 stdout: virusAligned.filtered.sortedByCoord.out.bam

tool/samtools/bam_filter_polyx.sh

@@ -1,5 +1,5 @@
 #!/bin/sh
 
 samtools view -h ${1} | \
-grep -v "AAAAAAAAAAAAAAAAAAAA" | grep -v "TTTTTTTTTTTTTTTTTTTT" | \
+grep -v "AAAAAAAAAAAAAAAAAAAA" | grep -v "TTTTTTTTTTTTTTTTTTTT" | grep -v "TGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTG" | \
 samtools view -bS - 

workflow/VIRTUS.PE.cwl

@@ -3,19 +3,19 @@
 class: Workflow
 cwlVersion: v1.0
 id: VIRTUS.PE
-doc: VIRTUS v1.1
+doc: VIRTUS v1.2
 label: VIRTUS.PE
 $namespaces:
   sbg: 'https://www.sevenbridges.com/'
 inputs:
   - id: fastq2
     type: File
-    'sbg:x': -532.9534301757812
-    'sbg:y': -424.9623107910156
+    'sbg:x': -446.9178771972656
+    'sbg:y': -426.26885986328125
   - id: fastq1
     type: File
-    'sbg:x': -534.9113159179688
-    'sbg:y': -269
+    'sbg:x': -445.51458740234375
+    'sbg:y': -274.6721496582031
   - id: genomeDir_human
     type: Directory
     'sbg:x': -305
@@ -26,8 +26,8 @@ inputs:
     'sbg:y': -750
   - id: nthreads
     type: int?
-    'sbg:x': -366.3414611816406
-    'sbg:y': 303.09088134765625
+    'sbg:x': -386.4197082519531
+    'sbg:y': 242.87530517578125
   - id: genomeDir_virus
     type: Directory
     'sbg:x': 417.4296875
@@ -67,20 +67,20 @@ outputs:
     outputSource:
       - star_mapping_pe_human/aligned
     type: File
-    'sbg:x': 209.65237426757812
-    'sbg:y': 379.7947998046875
+    'sbg:x': 190.84915161132812
+    'sbg:y': 381.26885986328125
   - id: output_unmapped
     outputSource:
       - samtools_view/output
     type: File
-    'sbg:x': 412
-    'sbg:y': 218
+    'sbg:x': 340.177001953125
+    'sbg:y': 225.9409942626953
   - id: output_fq2
     outputSource:
       - bedtools_bamtofastq_pe/output_fq2
     type: File?
-    'sbg:x': 549.3675537109375
-    'sbg:y': -334.5509948730469
+    'sbg:x': 558.2530517578125
+    'sbg:y': -337.0843505859375
   - id: output_fq1
     outputSource:
       - bedtools_bamtofastq_pe/output_fq1
@@ -156,8 +156,8 @@ steps:
       - id: out_fastq1
       - id: out_fastq2
     run: ../tool/fastp/fastp-pe.cwl
-    'sbg:x': -302
-    'sbg:y': -343
+    'sbg:x': -271.3704528808594
+    'sbg:y': -358.8819885253906
   - id: star_mapping_pe_human
     in:
       - id: fq1
@@ -192,8 +192,6 @@ steps:
     in:
       - id: threads
         source: nthreads
-      - id: b
-        default: true
       - id: f
         default: 4
       - id: prefix
@@ -204,8 +202,8 @@ steps:
       - id: output
     run: ../tool/samtools/samtools-view.cwl
     label: samtools-view
-    'sbg:x': 199.102294921875
-    'sbg:y': 52.59203338623047
+    'sbg:x': 179.7147216796875
+    'sbg:y': 53.134429931640625
   - id: bedtools_bamtofastq_pe
     in:
       - id: input
@@ -219,14 +217,14 @@ steps:
       - id: output_fq2
     run: ../tool/bedtools/bedtools-bamtofastq-pe.cwl
     label: bedtools-bamtofastq-pe
-    'sbg:x': 406.1430358886719
-    'sbg:y': 28.673505783081055
+    'sbg:x': 315.8917236328125
+    'sbg:y': -33.99026870727539
   - id: star_mapping_pe_virus
     in:
       - id: fq1
-        source: bedtools_bamtofastq_pe/output_fq1
+        source: fastq_pair/fq1_paired
       - id: fq2
-        source: bedtools_bamtofastq_pe/output_fq2
+        source: fastq_pair/fq2_paired
       - id: genomeDir
         source: genomeDir_virus
       - id: nthreads
@@ -247,8 +245,8 @@ steps:
       - id: unmapped
     run: ../tool/star/star_mapping-pe/star_mapping-pe.cwl
     label: 'STAR mapping: running mapping jobs.'
-    'sbg:x': 681.9188232421875
-    'sbg:y': 132.9390411376953
+    'sbg:x': 753.9961547851562
+    'sbg:y': 133.0281219482422
   - id: mk_virus_count
     in:
       - id: virus_bam
@@ -310,6 +308,47 @@ steps:
     label: bam_filter_polyX
     'sbg:x': 825.17626953125
     'sbg:y': 513.4646606445312
+  - id: kz_filter_fq2
+    in:
+      - id: threshold
+        default: 0.1
+      - id: input_fq
+        source: bedtools_bamtofastq_pe/output_fq2
+      - id: output_fq
+        default: kz_2.fq
+    out:
+      - id: output
+    run: ../tool/kz_filter/kz_filter.cwl
+    label: kz-filter_fq2
+    'sbg:x': 460.7408752441406
+    'sbg:y': 27.032846450805664
+  - id: kz_filter_fq1
+    in:
+      - id: threshold
+        default: 0.1
+      - id: input_fq
+        source: bedtools_bamtofastq_pe/output_fq1
+      - id: output_fq
+        default: kz_1.fq
+    out:
+      - id: output
+    run: ../tool/kz_filter/kz_filter.cwl
+    label: kz-filter_fq1
+    'sbg:x': 462.7408752441406
+    'sbg:y': 147
+  - id: fastq_pair
+    in:
+      - id: fq1
+        source: kz_filter_fq1/output
+      - id: fq2
+        source: kz_filter_fq2/output
+    out:
+      - id: fq1_paired
+      - id: fq2_paired
+    run: ../tool/fastq_pair/fastq_pair.cwl
+    label: fastq_pair
+    'sbg:x': 577.2360229492188
+    'sbg:y': 93.92456817626953
 requirements: []
 'sbg:license': CC BY-NC 4.0
 'sbg:toolAuthor': Yoshiaki Yasumizu

workflow/VIRTUS.SE.cwl

@@ -3,7 +3,7 @@
 class: Workflow
 cwlVersion: v1.0
 id: VIRTUS.SE
-doc: VIRTUS v1.1
+doc: VIRTUS v1.2
 label: VIRTUS.SE
 $namespaces:
   sbg: 'https://www.sevenbridges.com/'
@@ -124,8 +124,6 @@ steps:
     in:
       - id: threads
         source: nthreads
-      - id: b
-        default: true
       - id: f
         default: 4
       - id: prefix
@@ -236,12 +234,12 @@ steps:
       - id: output_fq
     run: ../tool/bedtools/bedtools-bamtofastq-se.cwl
     label: bedtools-bamtofastq-pe
-    'sbg:x': 392.1208801269531
-    'sbg:y': -110.00155639648438
+    'sbg:x': 350.34454345703125
+    'sbg:y': -113.04795837402344
   - id: star_mapping_se_virus
     in:
       - id: fq
-        source: bedtools_bamtofastq_se/output_fq
+        source: kz_filter/output
       - id: genomeDir
         source: genomeDir_virus
       - id: nthreads
@@ -274,9 +272,25 @@ steps:
     label: bam_filter_polyX
     'sbg:x': 847.6024780273438
     'sbg:y': 366.4100036621094
-requirements: []
+  - id: kz_filter
+    in:
+      - id: threshold
+        default: 0.1
+      - id: input_fq
+        source: bedtools_bamtofastq_se/output_fq
+      - id: output_fq
+        default: kz.fq
+    out:
+      - id: output
+    run: ../tool/kz_filter/kz_filter.cwl
+    label: kz-filter
+    'sbg:x': 489.5152893066406
+    'sbg:y': 39.372711181640625
+requirements:
+  - class: InlineJavascriptRequirement
+  - class: StepInputExpressionRequirement
+'sbg:license': CC BY-NC 4.0
 'sbg:links':
   - id: 'https://github.com/yyoshiaki/VIRTUS'
     label: ''
 'sbg:toolAuthor': Yoshiaki Yasumizu
-'sbg:license': CC BY-NC 4.0

workflow/kz_filter_PE.sh

@@ -0,0 +1,10 @@
+kz -k 2 < unmapped_1.fq > kz_1.txt
+kz -k 2 < unmapped_2.fq > kz_2.txt
+
+SCRIPT_DIR=$(cd $(dirname $0); pwd)
+python3 $SCRIPT_DIR/kz_list_PE.py
+
+list=(`cat kz_filter_list_1.txt` `cat kz_filter_list_2.txt`)
+
+samtools view virusAligned.filtered.sortedByCoord.out.bam | egrep -v "`echo $(IFS="|"; echo "${list[*]}")`" | cut -f3 | sort | uniq -c  > virus_counts_kz.txt
+samtools view virusAligned.filtered.sortedByCoord.out.bam | egrep "`echo $(IFS="|"; echo "${list[*]}")`" > kz_removed.txt

workflow/kz_filter_SE.sh

@@ -0,0 +1,8 @@
+kz -k 2 < unmapped.fq > kz.txt
+
+SCRIPT_DIR=$(cd $(dirname $0); pwd)
+python3 $SCRIPT_DIR/kz_list_SE.py
+
+list=(`cat kz_filter_list.txt`)
+samtools view virusAligned.filtered.sortedByCoord.out.bam | egrep -v "`echo $(IFS="|"; echo "${list[*]}")`" | cut -f3 | sort | uniq -c  > virus_counts_kz.txt
+samtools view virusAligned.filtered.sortedByCoord.out.bam | egrep "`echo $(IFS="|"; echo "${list[*]}")`" > kz_removed.txt