analysisRNAseqGRCh38.xml

<configuration name='RNAseqAnalysisGRCh38' description='for hg38 reference genome, paired end RNA seq analysis using STAR, feature counts, kallisto, arriba, RNAseQC and QualiMap2'
               configurationType='analysis' class='de.dkfz.roddy.core.Analysis'
               workflowClass='de.dkfz.b080.co.rnaseqworkflow.RNAseqWorkflow'
               runtimeServiceClass='de.dkfz.b080.co.common.COProjectsRuntimeService'
               listOfUsedTools=""
               usedToolFolders="rnaseqworkflow,tools"
               imports="RNAseqAnalysis"
               cleanupScript="cleanupScript">

    <configurationvalues>
        <cvalue name='SPECIES' value='Homo_sapiens' type='string' />
        <cvalue name='REF_GENOME' value='GRCh38' type='string'/>
        <cvalue name='REF_GENOME_VERSION' value='GRCh38_decoy_ebv_phiX' type='string'/>
        <cvalue name='GENCODE_VERSION' value='gencode_v39_chr_patch_hapl_scaff' type='string' />

        <cvalue name='hg38BaseDirectory' value='/omics/odcf/reference_data/by-species/${SPECIES}/${REF_GENOME}/${REF_GENOME_VERSION}' type='path' />

        <cvalue name='hg38BaseDirectory_legacy' value='/omics/odcf/reference_data/legacy/ngs_share/assemblies/hg_GRCh38/' type='path'/>
        <cvalue name='hg38DatabaseDirectory_legacy' value='${hg38BaseDirectory_legacy}/databases' type='string'/>

        <cvalue name='databaseDirectory' value='${hg38DatabaseDirectory_legacy}' type='string'/>
        <cvalue name='indexDirectory' value='${hg38BaseDirectory}' type='string'/>

        <cvalue name='GENOME_FA' value='${hg38BaseDirectory}/genome.fa' type='string'/>
        <cvalue name='GENOME_GATK_INDEX' value='${GENOME_FA}' type='string'/>

        <cvalue name='GENE_MODELS' value='${databaseDirectory}/gencode/${REF_GENOME_VERSION}/${GENCODE_VERSION}/annotation.gtf' type='string'/>
        <cvalue name='GENE_MODELS_NOGENE' value='${databaseDirectory}/gencode/${REF_GENOME_VERSION}/${GENCODE_VERSION}/annotation.nogene.gtf' type='string' 
                description="Like GENE_MODELS, but without 'gene' features. Required for gencode update for hg38. Defaults (for hg19): $GENE_MODELS."/>

        <cvalue name='GENE_MODELS_EXCLUDE' value='${databaseDirectory}/gencode/${REF_GENOME_VERSION}/${GENCODE_VERSION}/annotation.chrXYMT.rRNA.gtf' type='string'
                description="Other than mitochondrial genes, there were no genes annotated as tRNA, so have removed it from the filename."/>
        <cvalue name='GENE_MODELS_DEXSEQ' value='${databaseDirectory}/gencode/${REF_GENOME_VERSION}/${GENCODE_VERSION}/annotation.dexseq.gff' type='string'/>

        <cvalue name='GENOME_STAR_INDEX_100' value='${indexDirectory}/STAR/${STAR_VERSION}/${GENCODE_VERSION}' type='path' description='The default index generated by REFMAKE is based on the --sjdbOverhang 100'/>
        <cvalue name='GENOME_KALLISTO_INDEX' value='${indexDirectory}/kallisto/${KALLISTO_VERSION}/${GENCODE_VERSION}/kallisto.index' type='path'/>

        <cvalue name='GENOME_STAR_INDEX' value='${GENOME_STAR_INDEX_100}' type='string'/>
        <cvalue name='STAR_PARAMS_BASIC'
                value='--sjdbOverhang 100 --runThreadN ${CORES} --outFileNamePrefix ${STAR_PREFIX}. --genomeDir ${GENOME_STAR_INDEX} --runRNGseed 1234 --outTmpDir ${SCRATCH}/${SAMPLE}_${pid}_STAR'
                type='string'/>
         <cvalue name='STAR_PARAMS_CHIMERIC' value='--chimSegmentMin 15 --chimScoreMin 1 --chimScoreJunctionNonGTAG 0 --chimJunctionOverhangMin 15  --chimSegmentReadGapMax 3 --alignSJstitchMismatchNmax 5 -1 5 5 --chimOutType Junctions SeparateSAMold' type='string'/>
        <cvalue name='ARRIBA_KNOWN_FUSIONS' value='${hg38BaseDirectory_legacy}/tools_data/arriba/known_fusions_hg38_GRCh38_v${ARRIBA_VERSION}.tsv.gz' type='string'/>
        <cvalue name='ARRIBA_TAGS' value='${ARRIBA_KNOWN_FUSIONS}' type='string'/>
        <cvalue name='ARRIBA_BLACKLIST' value='${hg38BaseDirectory_legacy}/tools_data/arriba/blacklist_hg38_GRCh38_v${ARRIBA_VERSION}.tsv.gz' type='string'/>
        <cvalue name='ARRIBA_PROTEIN_DOMAINS' value='${hg38BaseDirectory_legacy}/tools_data/arriba/protein_domains_hg38_GRCh38_v${ARRIBA_VERSION}.gff3' type='string'/>
        <cvalue name='ARRIBA_CYTOBANDS' value='${hg38BaseDirectory_legacy}/tools_data/arriba/cytobands_hg38_GRCh38_v${ARRIBA_VERSION}.tsv' type="string"/>

        <!-- Hg38 Finger printing-->
        <cvalue name="fingerprintingSitesFile_hg38"
                value="${databaseDirectory}/fingerprinting/methylCtools-fingerprint/snp138Common.n1000_hg38.vh20140318.bed"
                type="path"/>
        <cvalue name="fingerprintingSitesFile" value="${fingerprintingSitesFile_hg38}" type="path"
                description='BED file with the reference positions used for fingerprinting. Used by alignment workflow for fingerprinting individuals to discover swaps. Use runFingerprinting=true/false to toggle fingerprinting.'/>
    </configurationvalues>
</configuration>