Merge pull request #28 from yyoshiaki/v2.0.2

V2.0.2

.gitignore

@@ -11,3 +11,4 @@ wrapper/*
 !wrapper/scattermap.py
 .vscode/settings.json
 ._.DS_Store
+.vscode/launch.json

README.md

@@ -1,4 +1,4 @@
-# VIRTUS2 : VIRal Transcript Usage Sensor v2.0.1 <img src="https://github.com/yyoshiaki/VIRTUS/raw/master/img/VIRTUS.jpg" width="20%" align="right" />
+# VIRTUS2 : VIRal Transcript Usage Sensor v2.0.2 <img src="https://github.com/yyoshiaki/VIRTUS/raw/master/img/VIRTUS.jpg" width="20%" align="right" />
 
 **!!Note : We updated VIRTUS to version2. In this version, we removed the gene quantification step by Salmon and single virus mode, and added coverage on viral genomes to the result to focus on virus-wide exploration. If you want to use the single virus mode, visit [https://github.com/yyoshiaki/VIRTUS](https://github.com/yyoshiaki/VIRTUS)**
 
@@ -318,6 +318,7 @@ usage: VIRTUS_wrapper.py [-h] \
                         [--fastq] \
                         [--figsize FIGSIZE] \
                         [--th_cov TH_COV] [--th_rate TH_RATE] \
+                        [--singularity] \
                         input_path
 
 positional arguments:
@@ -337,6 +338,7 @@ optional arguments:
   --figsize FIGSIZE     (default:6,6)
   --th_cov TH_COV       threshold of max viral coverage to plot, test (default:10)
   --th_rate TH_RATE     threshold of max rate virus/human to plot, test (default:0.0001)
+  --singularity         run with singularity (default:False)
 ```
 example
 ```

tool/kz_filter/Dockerfile

@@ -1,4 +1,4 @@
 FROM conda/miniconda3
 USER root
 
-RUN conda install -y -c eclarke komplexity
+RUN conda install -y -c eclarke komplexity==0.3.6

workflow/VIRTUS.PE.localref.job.yaml

@@ -0,0 +1,17 @@
+fastq1:
+  class: File
+  path: >-
+    ../test/ERR3240275/ERR3240275_1.fastq.gz
+fastq2:
+  class: File
+  path: >-
+    ../test/ERR3240275/ERR3240275_2.fastq.gz
+genomeDir_human:
+  class: Directory
+  path: ../test/STAR_index_human_localref
+genomeDir_virus:
+  class: Directory
+  path: ../test/STAR_index_virus_localref
+nthreads: 40
+outFileNamePrefix_human: human
+filename_output: VIRTUS.localref.output.tsv

workflow/VIRTUS.SE.localref.job.yaml

@@ -0,0 +1,13 @@
+fastq:
+  class: File
+  path: >-
+    ../test/SRR8315715/SRR8315715_2.fastq.gz
+genomeDir_human:
+  class: Directory
+  path: ../test/STAR_index_human_localref
+genomeDir_virus:
+  class: Directory
+  path: ../test/STAR_index_virus_localref
+nthreads: 40
+outFileNamePrefix_human: human
+filename_output: VIRTUS.localref.output.tsv

workflow/createindex_localref.cwl

@@ -0,0 +1,122 @@
+#!/usr/bin/env cwltool
+
+class: Workflow
+cwlVersion: v1.0
+id: createindex
+doc: VIRTUS v2.0
+label: CreateIndex_localref
+$namespaces:
+  sbg: 'https://www.sevenbridges.com/'
+inputs:
+  - id: f_virus
+    type: File
+    'sbg:x': -1123.8369140625
+    'sbg:y': -782.5
+  - id: dir_name_STAR_virus
+    type: string
+    'sbg:x': -1126.984130859375
+    'sbg:y': -518.8207397460938
+  - id: f_human
+    type: File
+    'sbg:x': -1102
+    'sbg:y': -167
+  - id: dir_name_STAR_human
+    type: string
+    'sbg:x': -1108
+    'sbg:y': 115
+  - id: runThreadN
+    type: int?
+    'sbg:x': -1116
+    'sbg:y': -318
+outputs:
+  - id: starIndex_virus
+    outputSource:
+      - star_index_virus/starIndex
+    type: Directory
+    'sbg:x': -486.8369140625
+    'sbg:y': -606.5
+  - id: starIndex_human
+    outputSource:
+      - star_index_human/starIndex
+    type: Directory
+    'sbg:x': -491.496826171875
+    'sbg:y': -166.5
+steps:
+  - id: star_index_virus
+    in:
+      - id: genomeDir
+        source: mkdir_virus/created_directory
+      - id: genomeFastaFiles
+        source: f_virus
+      - id: runThreadN
+        source: runThreadN
+      - id: genomeSAindexNbases
+        default: 12
+    out:
+      - id: starIndex
+    run: ../tool/star/star_index/star_index.cwl
+    label: 'STAR genomeGenerate: Generating genome indexes.'
+    'sbg:x': -709.8369140625
+    'sbg:y': -605.5
+  - id: mkdir_virus
+    in:
+      - id: dir_name
+        default: STAR_index_virus
+        source: dir_name_STAR_virus
+    out:
+      - id: created_directory
+    run: ../tool/mkdir/mkdir.cwl
+    label: Make directory if not exists
+    'sbg:x': -937.1593627929688
+    'sbg:y': -520.8286743164062
+
+  - id: star_index_human
+    in:
+      - id: genomeDir
+        source: mkdir_human/created_directory
+      - id: genomeFastaFiles
+        source: f_human
+      - id: runThreadN
+        source: runThreadN
+      - id: genomeSAindexNbases
+        default: 14
+    out:
+      - id: starIndex
+    run: ../tool/star/star_index/star_index.cwl
+    label: 'STAR genomeGenerate: Generating genome indexes.'
+    'sbg:x': -709.8369140625
+    'sbg:y': -605.5
+
+  - id: mkdir_human
+    in:
+      - id: dir_name
+        default: STAR_index_human
+        source: dir_name_STAR_human
+    out:
+      - id: created_directory
+    run: ../tool/mkdir/mkdir.cwl
+    label: Make directory if not exists
+    'sbg:x': -937.1593627929688
+    'sbg:y': -520.8286743164062
+
+
+  # - id: rnaseq_star_index_human
+  #   in:
+  #     - id: runThreadN
+  #       source: runThreadN
+  #     - id: url_genomefasta
+  #       source: url_genomefasta_human
+  #     - id: output_name_genomefasta
+  #       source: output_name_genomefasta_human
+  #     - id: dir_name
+  #       source: dir_name_STAR_human
+  #     - id: genomeSAindexNbases
+  #       default: 14
+  #   out:
+  #     - id: downloaded_genomefasta
+  #     - id: starIndex
+  #   run: rnaseq-star_index/rnaseq-star_index.cwl
+  #   'sbg:x': -688
+  #   'sbg:y': -91
+requirements:
+  - class: SubworkflowFeatureRequirement

workflow/createindex_localref.job.yaml

@@ -0,0 +1,11 @@
+dir_name_STAR_virus: STAR_index_virus_localref
+dir_name_STAR_human: STAR_index_human_localref
+runThreadN: 40
+f_virus:
+  class: File
+  path: >-
+    ../data/viruses.fasta
+f_human:
+  class: File
+  path: >- 
+    ../test/GRCh38.p13.genome.fa

wrapper/VIRTUS_wrapper.py

@@ -34,6 +34,7 @@
 parser.add_argument('--figsize', default = '8,3', help='(default:8,3)')
 parser.add_argument('--th_cov', default = '10', help='threshold of max viral coverage to plot, test (default:10)')
 parser.add_argument('--th_rate', default = '0.0001', help='threshold of max rate virus/human to plot, test (default:0.0001)')
+parser.add_argument('--singularity', action = 'store_true', help='run with singularity (default:False)')
 
 args = parser.parse_args()
 
@@ -53,6 +54,11 @@
 except (ValueError, IndexError):
     exit('invalid path to VIRTUS. try to change --VIRTUSDir to the absolute path.')
 
+if args.singularity:
+    singularity = ' --singularity'
+else:
+    singularity = ''
+
 # %%
 list_df_res = []
 
@@ -118,6 +124,7 @@
     if item["Layout"] =="PE":
         VIRTUS_cmd = " ".join([
             "cwltool",
+            singularity,
             "--rm-tmpdir",
             os.path.join(dir_VIRTUS, "VIRTUS.PE.cwl"), 
             "--fastq1", '../'+fastq1,
@@ -131,6 +138,7 @@
     elif item["Layout"] =="SE":
         VIRTUS_cmd = " ".join([
             "cwltool",
+            singularity,
             "--rm-tmpdir",
             os.path.join(dir_VIRTUS, "VIRTUS.SE.cwl"), 
             "--fastq", '../'+fastq,
@@ -195,8 +203,8 @@
 
 list_index = (set(list(df_rate.index)) & set(list(df_cov.index)))
 
-df_cov = df_cov.loc[list_index]
-df_rate = df_rate.loc[list_index]
+df_cov = df_cov.reindex(list_index)
+df_rate = df_rate.reindex(list_index)
 
 df_cov = df_cov[df_rate.columns]
 

wrapper/summary.csv

@@ -1,11 +1,11 @@
 virus,Ctrl_1_cov,Ctrl_2_cov,Flu_1_cov,Flu_2_cov,Ctrl_1_rate,Ctrl_2_rate,Flu_1_rate,Flu_2_rate,u-value,p-value,FDR
-"NC_007360.1_Influenza_A_virus_(A/Goose/Guangdong/1/96(H5N1))_nucleocapsid_protein_(NP)_gene,_complete_cds",7.53994,18.7859,79.0415,52.7796,2.0898752524203576e-08,2.0557772102157603e-07,0.00017833350419595113,5.705166243157474e-05,4.0,0.2452781168067728,0.27253124089641423
-"NC_007372.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_2,_complete_sequence",0.0,66.1256,99.5728,99.4874,0.0,7.537849770791118e-07,0.0001770468108072655,0.0001739883316582788,4.0,0.2452781168067728,0.27253124089641423
-"NC_001501.1_Moloney_murine_leukemia_virus,_complete_genome",0.0,99.808,0.0,99.808,0.0,0.00010224065325473046,0.0,0.00010721235153450432,2.5,1.0,1.0
-"NC_007366.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_4,_complete_sequence",0.0,88.6493,99.546,99.319,0.0,2.7752992337912757e-06,0.001292302052174779,0.001175295727694479,4.0,0.2452781168067728,0.27253124089641423
-"NC_004907.1_Influenza_A_virus_(A/Hong_Kong/1073/99(H9N2))_segment_7,_complete_sequence",0.0,26.0488,90.2439,64.7805,0.0,3.08366581532364e-07,0.00017954321592890345,0.00011777617866775727,4.0,0.2452781168067728,0.27253124089641423
-"NC_007369.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_5,_complete_sequence",0.0,95.6577,99.4891,99.3614,0.0,3.3920323968560035e-06,0.0015065639948394136,0.0014216742588635755,4.0,0.2452781168067728,0.27253124089641423
-"NC_007368.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_6,_complete_sequence",8.86162,95.8419,99.0457,98.9775,2.0898752524203576e-08,1.130677465618668e-06,0.0005835539425968027,0.0005348199832909638,4.0,0.2452781168067728,0.27253124089641423
-"NC_007370.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_8,_complete_sequence",0.0,96.0674,98.5393,98.3146,0.0,1.7131476751798e-06,0.0008950437164526317,0.0008666710694272907,4.0,0.2452781168067728,0.27253124089641423
-"NC_002016.1_Influenza_A_virus_(A/Puerto_Rico/8/1934(H1N1))_segment_7,_complete_sequence",0.0,31.4508,98.5394,97.7605,0.0,3.4262953503596e-07,0.00010884106382753628,0.0001308585341241699,4.0,0.2452781168067728,0.27253124089641423
-"NC_007367.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_7,_complete_sequence",0.0,97.7605,99.3184,99.0263,0.0,2.192829024230144e-06,0.0011559565424916634,0.0011260270173726686,4.0,0.2452781168067728,0.27253124089641423
+"NC_007368.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_6,_complete_sequence",8.86162,95.8419,99.0457,98.9775,2.0898752524203576e-08,1.130677465618668e-06,0.0005835539425968,0.0005348199832909,4.0,0.2452781168067728,0.27253124089641423
+"NC_002016.1_Influenza_A_virus_(A/Puerto_Rico/8/1934(H1N1))_segment_7,_complete_sequence",0.0,31.4508,98.5394,97.7605,0.0,3.4262953503596e-07,0.0001088410638275,0.0001308585341241,4.0,0.2452781168067728,0.27253124089641423
+"NC_007360.1_Influenza_A_virus_(A/Goose/Guangdong/1/96(H5N1))_nucleocapsid_protein_(NP)_gene,_complete_cds",7.53994,18.7859,79.0415,52.7796,2.0898752524203576e-08,2.05577721021576e-07,0.0001783335041959,5.7051662431574736e-05,4.0,0.2452781168067728,0.27253124089641423
+"NC_007370.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_8,_complete_sequence",0.0,96.0674,98.5393,98.3146,0.0,1.7131476751798e-06,0.0008950437164526,0.0008666710694272,4.0,0.2452781168067728,0.27253124089641423
+"NC_004907.1_Influenza_A_virus_(A/Hong_Kong/1073/99(H9N2))_segment_7,_complete_sequence",0.0,26.0488,90.2439,64.7805,0.0,3.08366581532364e-07,0.0001795432159289,0.0001177761786677,4.0,0.2452781168067728,0.27253124089641423
+"NC_007367.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_7,_complete_sequence",0.0,97.7605,99.3184,99.0263,0.0,2.192829024230144e-06,0.0011559565424916,0.0011260270173726,4.0,0.2452781168067728,0.27253124089641423
+"NC_007366.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_4,_complete_sequence",0.0,88.6493,99.546,99.319,0.0,2.775299233791276e-06,0.0012923020521747,0.0011752957276944,4.0,0.2452781168067728,0.27253124089641423
+"NC_001501.1_Moloney_murine_leukemia_virus,_complete_genome",0.0,99.808,0.0,99.808,0.0,0.0001022406532547,0.0,0.0001072123515345,2.5,1.0,1.0
+"NC_007369.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_5,_complete_sequence",0.0,95.6577,99.4891,99.3614,0.0,3.392032396856004e-06,0.0015065639948394,0.0014216742588635,4.0,0.2452781168067728,0.27253124089641423
+"NC_007372.1_Influenza_A_virus_(A/New_York/392/2004(H3N2))_segment_2,_complete_sequence",0.0,66.1256,99.5728,99.4874,0.0,7.537849770791119e-07,0.0001770468108072,0.0001739883316582,4.0,0.2452781168067728,0.27253124089641423