Replace species inference with seqwho

.gitlab-ci.yml

@@ -123,7 +123,7 @@ parseMetadata:
       - version_python.txt
     expire_in: 7 days
 
-inferMetadata:
+fastqc:
   stage: unit
   only:
     - push
@@ -132,22 +132,37 @@ inferMetadata:
     - merge_requests
     - schedules
   script:
-  - singularity run 'docker://gudmaprbk/rseqc4.0.0:1.0.0' infer_experiment.py --version > version_rseqc.txt
-  - >
-    align=$(echo $(grep "Overall alignment rate" ./test_data/meta/Q-Y5F6_1M.se.alignSummary.txt | cut -f2 -d ':' | cut -f2 -d ' ' | tr -d '%')) &&
-    if [[ ${align} == "" ]]; then exit 1; fi
-  - >
-    singularity run 'docker://gudmaprbk/rseqc4.0.0:1.0.0' infer_experiment.py -r "/project/BICF/BICF_Core/shared/gudmap/references/new/GRCh38.p13.v36/data/annotation/genome.bed" -i "./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam" 1>> Q-Y5F6_1M.se.inferMetadata.log &&
-    ended=`singularity run 'gudmaprbk/python3:1.0.0' python3 ./workflow/scripts/infer_meta.sh endness Q-Y5F6_1M.se.inferMetadata.log` &&
-    if [[ ${ended} == "" ]]; then exit 1; fi
-  - pytest -m inferMetadata
+  - singularity run 'docker://gudmaprbk/fastqc0.11.9:1.0.0' fastqc --version > version_fastqc.txt
+  - singularity run 'docker://gudmaprbk/fastqc0.11.9:1.0.0' fastqc ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz -o .
+  - pytest -m fastqc
+  artifacts:
+    name: "$CI_JOB_NAME"
+    when: always
+    paths:
+      - version_fastqc.txt
+    expire_in: 7 days
+
+seqwho:
+  stage: unit
+  only:
+    - push
+    - tags
+  except:
+    - merge_requests
+    - schedules
+  script:
+  - wget -O SeqWho.ix https://cloud.biohpc.swmed.edu/index.php/s/eeNWqZz8jqN5zWY/download
+  - singularity run 'docker://gudmaprbk/seqwho0.0.1:1.0.0' seqwho.py | grep -o Version.* > version_seqwho.txt
+  - singularity run 'docker://gudmaprbk/seqwho0.0.1:1.0.0' seqwho.py -f ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz -x SeqWho.ix
+  - pytest -m seqwho
   artifacts:
     name: "$CI_JOB_NAME"
     when: always
     paths:
-      - version_rseqc.txt
+      - version_seqwho.txt
     expire_in: 7 days
 
+
 trimData:
   stage: unit
   only:
@@ -182,6 +197,31 @@ downsampleData:
   - singularity run 'docker://gudmaprbk/seqtk1.3:1.0.0' seqtk sample -s100 ./test_data/fastq/small/Q-Y5F6_1M.se_trimmed.fq.gz 1000 1> sampled.1.fq
   - pytest -m downsampleData
 
+  inferMetadata:
+    stage: unit
+    only:
+      - push
+      - tags
+    except:
+      - merge_requests
+      - schedules
+    script:
+    - singularity run 'docker://gudmaprbk/rseqc4.0.0:1.0.0' infer_experiment.py --version > version_rseqc.txt
+    - >
+      align=$(echo $(grep "Overall alignment rate" ./test_data/meta/Q-Y5F6_1M.se.alignSummary.txt | cut -f2 -d ':' | cut -f2 -d ' ' | tr -d '%')) &&
+      if [[ ${align} == "" ]]; then exit 1; fi
+    - >
+      singularity run 'docker://gudmaprbk/rseqc4.0.0:1.0.0' infer_experiment.py -r "/project/BICF/BICF_Core/shared/gudmap/references/new/GRCh38.p13.v36/data/annotation/genome.bed" -i "./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam" 1>> Q-Y5F6_1M.se.inferMetadata.log &&
+      ended=`singularity run 'gudmaprbk/python3:1.0.0' python3 ./workflow/scripts/infer_meta.sh endness Q-Y5F6_1M.se.inferMetadata.log` &&
+      if [[ ${ended} == "" ]]; then exit 1; fi
+    - pytest -m inferMetadata
+    artifacts:
+      name: "$CI_JOB_NAME"
+      when: always
+      paths:
+        - version_rseqc.txt
+      expire_in: 7 days
+  
 alignData:
   stage: unit
   only:
@@ -282,26 +322,6 @@ makeBigWig:
       - version_deeptools.txt
     expire_in: 7 days
 
-fastqc:
-  stage: unit
-  only:
-    - push
-    - tags
-  except:
-    - merge_requests
-    - schedules
-  script:
-  - singularity run 'docker://gudmaprbk/fastqc0.11.9:1.0.0' fastqc --version > version_fastqc.txt
-  - singularity run 'docker://gudmaprbk/fastqc0.11.9:1.0.0' fastqc ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz -o .
-  - pytest -m fastqc
-  artifacts:
-    name: "$CI_JOB_NAME"
-    when: always
-    paths:
-      - version_fastqc.txt
-    expire_in: 7 days
-
-
 dataQC:
   stage: unit
   only:

docs/software_references_mqc.yaml

@@ -80,12 +80,18 @@
                 <li>FastQC <a href="https://www.bioinformatics.babraham.ac.uk/projects/fastqc/" class="uri">https://www.bioinformatics.babraham.ac.uk/projects/fastqc/</a></li>
                 </ul>
                 <ol start="13" style="list-style-type: decimal">
+                <li><strong>SeqWho</strong></li>
+                </ol>
+                <ul>
+                <li>SeqWho <a href="https://git.biohpc.swmed.edu/s181649/seqwho" class="uri">https://git.biohpc.swmed.edu/s181649/seqwho/</a></li>
+                </ul>
+                <ol start="14" style="list-style-type: decimal">
                 <li><strong>MultiQC</strong>:</li>
                 </ol>
                 <ul>
                 <li>Ewels P., Magnusson M., Lundin S. and Käller M. 2016. MultiQC: Summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32(19): 3047–3048. doi:<a href="https://dx.doi.org/10.1093/bioinformatics/btw354">10.1093/bioinformatics/btw354</a></li>
                 </ul>
-                <ol start="14" style="list-style-type: decimal">
+                <ol start="15" style="list-style-type: decimal">
                 <li><strong>Nextflow</strong>:</li>
                 </ol>
                 <ul>

workflow/conf/aws.config

@@ -32,6 +32,18 @@ process {
     cpus = 15
     memory = '1 GB'
   }
+  withName:getRefERCC {
+    cpus = 1
+    memory = '1 GB'
+  }
+  withName:getRef {
+    cpus = 1
+    memory = '1 GB'
+  }
+  withName:fastqc {
+    cpus = 1
+    memory = '1 GB'
+  }
   withName:seqwho {
     cpus = 1
     memory = '1 GB'
@@ -40,14 +52,14 @@ process {
     cpus = 20
     memory = '2 GB'
   }
-  withName:getRefInfer {
-    cpus = 1
-    memory = '1 GB'
-  }
   withName:downsampleData {
     cpus = 1
     memory = '1 GB'
   }
+  withName:alignSampleDataERCC {
+    cpus = 50
+    memory = '5 GB'
+  }
   withName:alignSampleData {
     cpus = 50
     memory = '5 GB'
@@ -60,10 +72,6 @@ process {
     cpus = 1
     memory = '1 GB'
   }
-  withName:getRef {
-    cpus = 1
-    memory = '1 GB'
-  }
   withName:alignData {
     cpus = 50
     memory = '10 GB'
@@ -80,10 +88,6 @@ process {
     cpus = 15
     memory = '5 GB'
   }
-  withName:fastqc {
-    cpus = 1
-    memory = '1 GB'
-  }
   withName:dataQC {
     cpus = 15
     memory = '2 GB'

workflow/conf/biohpc.config

@@ -22,18 +22,27 @@ process {
   withName:parseMetadata {
     executor = 'local'
   }
+  withName:getRefERCC {
+    queue = 'super'
+  }
+  withName:getRef {
+    queue = 'super'
+  }
+  withName:fastqc {
+    queue = 'super'
+  }
   withName:seqwho {
     executor = 'local'
   }
   withName:trimData {
     queue = 'super'
   }
-  withName:getRefInfer {
-    queue = 'super'
-  }
   withName:downsampleData {
     executor = 'local'
   }
+  withName:alignSampleDataERCC {
+    queue = '128GB,256GB,256GBv1,384GB'
+  }
   withName:alignSampleData {
     queue = '128GB,256GB,256GBv1,384GB'
   }
@@ -43,9 +52,6 @@ process {
   withName:checkMetadata {
     executor = 'local'
   }
-  withName:getRef {
-    queue = 'super'
-  }
   withName:alignData {
     queue = '256GB,256GBv1'
   }
@@ -58,9 +64,6 @@ process {
   withName:makeBigWig {
     queue = 'super'
   }
-  withName:fastqc {
-    queue = 'super'
-  }
   withName:dataQC {
     queue = 'super'
   }

workflow/nextflow.config

@@ -28,18 +28,27 @@ process {
   withName:parseMetadata {
     container = 'gudmaprbk/python3:1.0.0'
   }
+  withName:getRefERCC {
+    container = 'gudmaprbk/deriva1.4:1.0.0'
+  }
+  withName:getRef {
+    container = 'gudmaprbk/deriva1.4:1.0.0'
+  }
+  withName:fastqc {
+    container = 'gudmaprbk/fastqc0.11.9:1.0.0'
+  }
   withName:seqwho {
     container = 'gudmaprbk/seqwho0.0.1:1.0.0'
   }
   withName:trimData {
     container = 'gudmaprbk/trimgalore0.6.5:1.0.0'
   }
-  withName:getRefInfer {
-    container = 'gudmaprbk/deriva1.4:1.0.0'
-  }
   withName:downsampleData {
     container = 'gudmaprbk/seqtk1.3:1.0.0'
   }
+  withName:alignSampleDataERCC {
+    container = 'gudmaprbk/hisat2.2.1:1.0.0'
+  }
   withName:alignSampleData {
     container = 'gudmaprbk/hisat2.2.1:1.0.0'
   }
@@ -49,9 +58,6 @@ process {
   withName:checkMetadata {
     container = 'gudmaprbk/gudmap-rbk_base:1.0.0'
   }
-  withName:getRef {
-    container = 'gudmaprbk/deriva1.4:1.0.0'
-  }
   withName:alignData {
     container = 'gudmaprbk/hisat2.2.1:1.0.0'
   }
@@ -64,9 +70,6 @@ process {
   withName:makeBigWig {
     container = 'gudmaprbk/deeptools3.5.0:1.0.0'
   }
-  withName:fastqc {
-    container = 'gudmaprbk/fastqc0.11.9:1.0.0'
-  }
   withName:dataQC {
     container = 'gudmaprbk/rseqc4.0.0:1.0.0'
   }

workflow/scripts/generate_versions.py

@@ -34,6 +34,7 @@ SOFTWARE_REGEX = {
     'Python': ['version_python.txt', r"Python (\S+)"],
     'DERIVA': ['version_deriva.txt', r"(\S+)"],
     'BDBag': ['version_bdbag.txt', r"BDBag (\S+) \(Bagit \S+\)"],
+    'SeqWho': ['version_seqwho.txt', r"Version: (\S+)"],
     'RSeQC': ['version_rseqc.txt', r"infer_experiment.py (\S+)"],
     'Trim Galore!': ['version_trimgalore.txt', r"version (\S+)"],
     'HISAT2': ['version_hisat2.txt', r"version (\S+)"],
@@ -93,6 +94,7 @@ def main():
     results['Python'] = '<span style="color:#999999;\">Not Run</span>'
     results['DERIVA'] = '<span style="color:#999999;\">Not Run</span>'
     results['BDBag'] = '<span style="color:#999999;\">Not Run</span>'
+    results['SeqWho'] = '<span style="color:#999999;\">Not Run</span>'
     results['RSeQC'] = '<span style="color:#999999;\">Not Run</span>'
     results['Trim Galore!'] = '<span style="color:#999999;\">Not Run</span>'
     results['HISAT2'] = '<span style="color:#999999;\">Not Run</span>'

workflow/scripts/get_updated_badge_info.sh

@@ -13,6 +13,7 @@ echo "collecting tool version for badges"
 python_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o Python.* | grep -oP "(?<=d>).*(?=\<)")
 deriva_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o DERIVA.* | grep -oP "(?<=d>).*(?=\<)")
 bdbag_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o BDBag.* | grep -oP "(?<=d>).*(?=\<)")
+seqwho_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o SeqWho.* | grep -oP "(?<=d>).*(?=\<)")
 rseqc_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o RSeQC.* | grep -oP "(?<=d>).*(?=\<)")
 trimgalore_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o 'Trim Galore!'.* | grep -oP "(?<=d>).*(?=\<)")
 hisat2_version=$(git show ${latest_release_tag}:docs/software_versions_mqc.yaml | grep -o HISAT2.* | grep -oP "(?<=d>).*(?=\<)")
@@ -36,6 +37,7 @@ curl --request GET https://img.shields.io/badge/Nextflow%20Version-${develop_nex
 
 curl --request GET https://img.shields.io/badge/Python%20Version-${python_version}-blueviolet?style=flat > ./badges/tools/python.svg
 curl --request GET https://img.shields.io/badge/DERIVA%20Version-${deriva_version}-blueviolet?style=flat > ./badges/tools/deriva.svg
+curl --request GET https://img.shields.io/badge/SeqWho%20Version-${seqwho_version}-blueviolet?style=flat > ./badges/tools/seqwho.svg
 curl --request GET https://img.shields.io/badge/BDBag%20Version-${bdbag_version}-blueviolet?style=flat > ./badges/tools/bdbag.svg
 curl --request GET https://img.shields.io/badge/RSeQC%20Version-${rseqc_version}-blueviolet?style=flat > ./badges/tools/rseqc.svg
 curl --request GET https://img.shields.io/badge/Trim%20Galore%20Version-${trimgalore_version}-blueviolet?style=flat > ./badges/tools/trimgalore.svg

workflow/tests/test_seqwho.py

+#!/usr/bin/env python3
+
+import pytest
+import pandas as pd
+from io import StringIO
+import os
+
+test_output_path = os.path.dirname(os.path.abspath(__file__)) + \
+    '/../../'
+
+
+@pytest.mark.seqwho
+def test_seqwho():
+    assert os.path.exists(os.path.join(
+        test_output_path, 'SeqWho_call.tsv'))