.gitlab-ci.yml

before_script:
  - module load python/3.6.4-anaconda
  - pip install --user  attrs==19.1.0 pytest-pythonpath==0.7.1 pytest-cov==2.5.1 deriva==1.3.0
  - module load singularity/3.5.3
  - module load nextflow/20.01.0
  - ln -sfn /project/BICF/BICF_Core/shared/gudmap/test_data/* ./test_data/
  - mkdir -p ~/.deriva
  - mkdir -p ~/.bdbag

stages:
  - unit
  - reference
  - integration
  - consistency

getBag:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - ln -sfn `readlink -e ./test_data/auth/credential.json` ~/.deriva/credential.json
  - singularity run 'docker://bicf/gudmaprbkfilexfer:2.0.1_indev' deriva-download-cli dev.gudmap.org --catalog 2 ./workflow/conf/replicate_export_config.json . rid=Q-Y5F6
  - pytest -m getBag

getData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - ln -sfn `readlink -e ./test_data/auth/cookies.txt` ~/.bdbag/deriva-cookies.txt
  - unzip ./test_data/bag/Replicate_Q-Y5F6.zip
  - singularity run 'docker://bicf/gudmaprbkfilexfer:2.0.1_indev' bash ./workflow/scripts/bdbagFetch.sh Replicate_Q-Y5F6 Replicate_Q-Y5F6 TEST
  - pytest -m getData

parseMetadata:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - rep=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p repRID)
  - exp=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p expRID)
  - study=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p studyRID)
  - endsMeta=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p endsMeta)
  - endsManual=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p endsManual)
  - stranded=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p stranded)
  - spike=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p spike)
  - species=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.csv" -p species)
  - readLength=$(singularity run 'docker://bicf/python3:2.0.1_indev' python3 ./workflow/scripts/parseMeta.py -r Replicate_RID -m "./test_data/meta/metaTest.stageNew.csv" -p readLength)
  - echo -e "${endsMeta},${endsManual},${stranded},${spike},${species},${readLength},${exp},${study},${rep}" > design.csv
  - pytest -m parseMetadata

inferMetadata:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - >
    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://bicf/rseqc3.0:2.0.1_indev' infer_experiment.py -r "/project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/bed/genome.bed" -i "./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam" 1>> Q-Y5F6_1M.se.inferMetadata.log &&
    ended=`singularity run 'docker://bicf/python3:1.3' python3 ./workflow/scripts/inferMeta.sh endness Q-Y5F6_1M.se.inferMetadata.log` &&
    if [[ ${ended} == "" ]]; then exit 1; fi
  - pytest -m inferMetadata

trimData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/trimgalore:1.1' trim_galore --gzip -q 25 --length 35 --basename Q-Y5F6_1M.se ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz
  - singularity run 'docker://bicf/trimgalore:1.1' trim_galore --gzip -q 25 --length 35 --paired --basename Q-Y5F6_1M.pe ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz ./test_data/fastq/small/Q-Y5F6_1M.R2.fastq.gz
  - readLengthSE=$(zcat *_trimmed.fq.gz | awk '{if(NR%4==2) print length($1)}' | sort -n | awk '{a[NR]=$0}END{print(NR%2==1)?a[int(NR/2)+1]:(a[NR/2]+a[NR/2+1])/2}')
  - readLengthPE=$(zcat *_1.fq.gz | awk '{if(NR%4==2) print length($1)}' | sort -n | awk '{a[NR]=$0}END{print(NR%2==1)?a[int(NR/2)+1]:(a[NR/2]+a[NR/2+1])/2}')
  - pytest -m trimData

downsampleData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/seqtk:2.0.1_indev' seqtk sample -s100 ./test_data/fastq/small/Q-Y5F6_1M.se_trimmed.fq.gz 1000 1> sampled.1.fq
  - pytest -m downsampleData

alignData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' hisat2 -p 20 --add-chrname --un-gz Q-Y5F6_1M.se.unal.gz -S Q-Y5F6_1M.se.sam -x /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/hisat2/genome --rna-strandness F -U ./test_data/fastq/small/Q-Y5F6_1M.se_trimmed.fq.gz --summary-file Q-Y5F6_1M.se.alignSummary.txt --new-summary
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools view -1 -@ 20 -F 4 -F 8 -F 256 -o Q-Y5F6_1M.se.bam Q-Y5F6_1M.se.sam
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools sort -@ 20 -O BAM -o Q-Y5F6_1M.se.sorted.bam Q-Y5F6_1M.se.bam
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools index -@ 20 -b Q-Y5F6_1M.se.sorted.bam Q-Y5F6_1M.se.sorted.bam.bai
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' hisat2 -p 20 --add-chrname --un-gz Q-Y5F6_1M.pe.unal.gz -S Q-Y5F6_1M.pe.sam -x /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/hisat2/genome --rna-strandness FR --no-mixed --no-discordant -1 ./test_data/fastq/small/Q-Y5F6_1M.pe_R1_val_1.fq.gz -2 ./test_data/fastq/small/Q-Y5F6_1M.pe_R2_val_2.fq.gz --summary-file Q-Y5F6_1M.pe.alignSummary.txt --new-summary
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools view -1 -@ 20 -F 4 -F 8 -F 256 -o Q-Y5F6_1M.pe.bam Q-Y5F6_1M.pe.sam
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools sort -@ 20 -O BAM -o Q-Y5F6_1M.pe.sorted.bam Q-Y5F6_1M.pe.bam
  - singularity run 'docker://bicf/gudmaprbkaligner:2.0.1_indev' samtools index -@ 20 -b Q-Y5F6_1M.pe.sorted.bam Q-Y5F6_1M.pe.sorted.bam.bai
  - pytest -m alignData

dedupData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/gudmaprbkdedup:2.0.0' java -jar /picard/build/libs/picard.jar MarkDuplicates I=./test_data/bam/small/Q-Y5F6_1M.se.sorted.bam O=Q-Y5F6_1M.se.deduped.bam M=Q-Y5F6_1M.se.deduped.Metrics.txt REMOVE_DUPLICATES=true
  - singularity run 'docker://bicf/gudmaprbkdedup:2.0.0' samtools sort -@ 20 -O BAM -o Q-Y5F6_1M.se.sorted.deduped.bam ./test_data/bam/small/Q-Y5F6_1M.se.deduped.bam
  - singularity run 'docker://bicf/gudmaprbkdedup:2.0.0' samtools index -@ 20 -b ./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam Q-Y5F6_1M.se.sorted.deduped.bam.bai
  - >
    for i in {"chr8","chr4","chrY"}; do
      echo "samtools view -b Q-Y5F6_1M.se.sorted.deduped.bam ${i} > Q-Y5F6_1M.se.sorted.deduped.${i}.bam; samtools index -@ 20 -b Q-Y5F6_1M.se.sorted.deduped.${i}.bam Q-Y5F6_1M.se.sorted.deduped.${i}.bam.bai;";
      done | singularity run 'docker://bicf/gudmaprbkdedup:2.0.0' parallel -j 20 -k
  - pytest -m dedupData

countData:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - ln -s /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/geneID.tsv
  - ln -s /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/Entrez.tsv
  - singularity run 'docker://bicf/subread2:2.0.0' featureCounts -T 20 -a /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/genome.gtf -G /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/genome.fna -g 'gene_name' --extraAttributes 'gene_id' -o Q-Y5F6_1M.se.countData -s 1 -R SAM --primary --ignoreDup ./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam
  - singularity run 'docker://bicf/subread2:2.0.0' Rscript ./workflow/scripts/calculateTPM.R --count ./test_data/counts/small/Q-Y5F6_1M.se.countData
  - singularity run 'docker://bicf/subread2:2.0.0' Rscript ./workflow/scripts/convertGeneSymbols.R --repRID Q-Y5F6_1M.se
  - assignedReads=$(grep -m 1 'Assigned' *.summary | grep -oe '\([0-9.]*\)')
  - pytest -m makeFeatureCounts

makeBigWig:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/deeptools3.3:2.0.1_indev' bamCoverage -p 20 -b ./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.bam -o Q-Y5F6_1M.se.bw
  - pytest -m makeBigWig

fastqc:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - singularity run 'docker://bicf/fastqc:2.0.1_indev' fastqc ./test_data/fastq/small/Q-Y5F6_1M.R1.fastq.gz -o .
  - pytest -m fastqc

dataQC:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - echo -e  "geneID\tchrom\ttx_start\ttx_end\tTIN" > Q-Y5F6_1M.se.sorted.deduped.tin.xls
  - for i in {"chr8","chr4","chrY"}; do
    echo "tin.py -i ./test_data/bam/small/Q-Y5F6_1M.se.sorted.deduped.${i}.bam -r /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/bed/genome.bed; cat Q-Y5F6_1M.se.sorted.deduped.${i}.tin.xls | tr -s \"\\w\" \"\\t\" | grep -P \"\\t${i}\\t\";"; done | singularity run 'docker://bicf/rseqc3.0:2.0.1_indev' parallel -j 20 -k >> Q-Y5F6_1M.se.sorted.deduped.tin.xls
  - pytest -m dataQC

outputBag:
  stage: unit
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - mkdir test
  - singularity run 'docker://bicf/gudmaprbkfilexfer:2.0.1_indev' bdbag test --archiver zip
  - pytest -m outputBag


humanBioHPC:
  stage: reference
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - mkdir -p hu
  - cp -R /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/hisat2 ./hu/

mouseBioHPC:
  stage: reference
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - mkdir -p mo
  - cp -R /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/hisat2 ./mo/

humanDataHub:
  stage: reference
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - ln -sfn `readlink -e ./test_data/auth/cookies.txt` ~/.bdbag/deriva-cookies.txt
  - referenceBase=dev.gudmap.org
  - refName=GRCh
  - refHuVersion=38.p12.v31
  - references=$(echo ${referenceBase}/${refName}${refHuVersion})
  - GRCv=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f1)
  - GRCp=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f2)
  - GENCODE=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f3)
  - query=$(echo 'https://'${referenceBase}'/ermrest/catalog/2/entity/RNASeq:Reference_Genome/Reference_Version='${GRCv}'.'${GRCp}'/Annotation_Version=GENCODE%20'${GENCODE})
  - curl --request GET ${query} > refQuery.json
  - refURL=$(python ./workflow/scripts/extractRefData.py --returnParam URL)
  - loc=$(dirname ${refURL})
  - if [ "${loc}" = "/hatrac/*" ]; then echo "reference not present in hatrac"; exit 1; fi
  - filename=$(echo $(basename ${refURL}) | grep -oP '.*(?=:)')
  - test=$(singularity run 'docker://gudmaprbk/deriva1.3:1.0.0' deriva-hatrac-cli --host ${referenceBase} ls ${loc}/)
  - test=$(echo ${test} | grep -o ${filename})
  - if [ "${test}" == "" ]; then echo "reference file not present"; exit 1; fi

mousenDataHub:
  stage: reference
  only:
    - push
    - tags
  except:
    - merge_requests
  script:
  - ln -sfn `readlink -e ./test_data/auth/cookies.txt` ~/.bdbag/deriva-cookies.txt
  - referenceBase=dev.gudmap.org
  - refName=GRCm
  - refHuVersion=38.p6.vM22
  - references=$(echo ${referenceBase}/${refName}${refHuVersion})
  - GRCv=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f1)
  - GRCp=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f2)
  - GENCODE=$(echo ${references} | grep -o ${refName}.* | cut -d '.' -f3)
  - query=$(echo 'https://'${referenceBase}'/ermrest/catalog/2/entity/RNASeq:Reference_Genome/Reference_Version='${GRCv}'.'${GRCp}'/Annotation_Version=GENCODE%20'${GENCODE})
  - curl --request GET ${query} > refQuery.json
  - refURL=$(python ./workflow/scripts/extractRefData.py --returnParam URL)
  - loc=$(dirname ${refURL})
  - if [ "${loc}" = "/hatrac/*" ]; then echo "reference not present in hatrac"; exit 1; fi
  - filename=$(echo $(basename ${refURL}) | grep -oP '.*(?=:)')
  - test=$(singularity run 'docker://gudmaprbk/deriva1.3:1.0.0' deriva-hatrac-cli --host ${referenceBase} ls ${loc}/)
  - test=$(echo ${test} | grep -o ${filename})
  - if [ "${test}" == "" ]; then echo "reference file not present"; exit 1; fi


integration_se:
  stage: integration
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - hostname
  - ulimit -a
  - nextflow -q run ./workflow/rna-seq.nf --deriva ./test_data/auth/credential.json --bdbag ./test_data/auth/cookies.txt --repRID 16-1ZX4 -with-dag dag.png --ci true
  - find . -type f -name "multiqc_data.json" -exec cp {} ./SE_multiqc_data.json \;
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - output/qc/
      - output/report/
      - SE_multiqc_data.json
    expire_in: 7 days
  retry:
    max: 1
    when:
      - always

integration_pe:
  stage: integration
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - hostname
  - ulimit -a
  - nextflow -q run ./workflow/rna-seq.nf --deriva ./test_data/auth/credential.json --bdbag ./test_data/auth/cookies.txt --repRID Q-Y5JA -with-dag dag.png --ci true
  - find . -type f -name "multiqc_data.json" -exec cp {} ./PE_multiqc_data.json \;
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - dag.png
      - output/qc/
      - output/report/
      - PE_multiqc_data.json
    expire_in: 7 days
  retry:
    max: 1
    when:
      - always

override_inputBag:
  stage: integration
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - hostname
  - ulimit -a
  - nextflow -q run ./workflow/rna-seq.nf --deriva ./test_data/auth/credential.json --bdbag ./test_data/auth/cookies.txt --repRID Q-Y5F6 --inputBagForce ./test_data/bag/Replicate_Q-Y5F6.zip --ci true
  - find . -type f -name "multiqc_data.json" -exec cp {} ./inputBagOverride_PE_multiqc_data.json \;
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - inputBagOverride_PE_multiqc_data.json
    expire_in: 7 days
  retry:
    max: 1
    when:
      - always

override_fastq:
  stage: integration
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - hostname
  - ulimit -a
  - nextflow -q run ./workflow/rna-seq.nf --deriva ./test_data/auth/credential.json --bdbag ./test_data/auth/cookies.txt --repRID Q-Y5F6 --fastqsForce './test_data/fastq/small/Q-Y5F6_1M.R{1,2}.fastq.gz' --ci true
  - find . -type f -name "multiqc_data.json" -exec cp {} ./fastqOverride_PE_multiqc_data.json \;
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - fastqOverride_PE_multiqc_data.json
    expire_in: 7 days
  retry:
    max: 1
    when:
      - always

override_species:
  stage: integration
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - hostname
  - ulimit -a
  - nextflow -q run ./workflow/rna-seq.nf --deriva ./test_data/auth/credential.json --bdbag ./test_data/auth/cookies.txt --repRID Q-Y5ER --speciesForce 'Homo sapiens' --ci true
  - find . -type f -name "multiqc_data.json" -exec cp {} ./speciesOverride_PE_multiqc_data.json \;
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - speciesOverride_PE_multiqc_data.json
    expire_in: 7 days
  retry:
    max: 1
    when:
      - always


consistency:
  stage: consistency
  only: [merge_requests]
  except:
    variables:
        - $CI_MERGE_REQUEST_TARGET_BRANCH_NAME =~ /master/
  script:
  - pytest -m consistencySE
  - pytest -m consistencyPE
  artifacts:
    name: "$CI_JOB_NAME"
    when: always
    paths:
      - SE_multiqc_data.json
      - PE_multiqc_data.json
    expire_in: 7 days