Merge pull request #243 from biowdl/bcftools_setid

Bcftools: added annotation and sorting

CHANGELOG.md

@@ -11,6 +11,7 @@ that users understand how the changes affect the new version.
 
 version 5.0.0-dev
 ---------------------------
++ bcftools: add sorting and annotation
 + Bam2fastx: Input bam and index are now arrays.
 + Lima: Remove globs from outputs.
 + Updated task gridss.wdl: add --jvmheap parameter

bcftools.wdl

@@ -22,6 +22,149 @@ version 1.0
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
+task Annotate {
+    input {
+        File? annsFile
+        String? collapse
+        Array[String] columns = []
+        String? exclude
+        Boolean force = false
+        File? headerLines
+        String? newId
+        String? include
+        Boolean keepSites = false
+        String? markSites
+        Boolean noVersion = false
+        String outputType = "z"
+        String? regions
+        File? regionsFile
+        File? renameChrs
+        Array[String] samples = []
+        File? samplesFile
+        Boolean singleOverlaps = false
+        Array[String] removeAnns = []
+        File inputFile
+        String outputPath = "output.vcf.gz"
+        
+        Int threads = 0
+        String memory = "256M"
+        Int timeMinutes = 1 + ceil(size(inputFile, "G"))
+        String dockerImage = "quay.io/biocontainers/bcftools:1.10.2--h4f4756c_2"
+    }
+
+    command {
+        set -e 
+        mkdir -p "$(dirname ~{outputPath})"
+        bcftools annotate \
+        -o ~{outputPath} \
+        -O ~{outputType} \
+        ~{"--annotations " + annsFile} \
+        ~{"--collapse " + collapse} \
+        ~{true="--columns" false="" length(columns) > 0} ~{sep="," columns} \
+        ~{"--exclude " + exclude} \
+        ~{true="--force" false="" force} \
+        ~{"--header-lines " + headerLines} \
+        ~{"--set-id " + newId} \
+        ~{"--include " + include} \
+        ~{true="--keep-sites" false="" keepSites} \
+        ~{"--mark-sites " + markSites} \
+        ~{true="--no-version" false="" noVersion} \
+        ~{"--regions " + regions} \
+        ~{"--regions-file " + regionsFile} \
+        ~{"--rename-chrs " + renameChrs} \
+        ~{true="--samples" false="" length(samples) > 0} ~{sep="," samples} \
+        ~{"--samples-file " + samplesFile} \
+        ~{true="--single-overlaps" false="" singleOverlaps} \
+        ~{true="--remove" false="" length(removeAnns) > 0} ~{sep="," removeAnns} \
+        ~{inputFile}
+        bcftools index --tbi ~{outputPath}
+
+    }
+    
+    output {
+        File outputVcf = outputPath
+        File outputVcfIndex = outputPath + ".tbi"
+    }
+
+    runtime {
+        memory: memory
+        time_minutes: timeMinutes
+        docker: dockerImage
+    }
+
+    parameter_meta {
+        outputPath: {description: "The location the output VCF file should be written.", category: "common"}
+        outputType: {description: "Output type: v=vcf, z=vcf.gz, b=bcf, u=uncompressed bcf", category: "advanced"}
+        annsFile: {description: "Bgzip-compressed and tabix-indexed file with annotations (see man page for details).", category: "advanced"}
+        collapse: {description: "Treat as identical records with <snps|indels|both|all|some|none>, see man page for details.", category: "advanced"}
+        columns: {description: "Comma-separated list of columns or tags to carry over from the annotation file (see man page for details).", category: "advanced"}
+        exclude: {description: "Exclude sites for which the expression is true (see man page for details).", category: "advanced"}
+        force: {description: "Continue even when parsing errors, such as undefined tags, are encountered.", category: "advanced"}
+        headerLines: {description: "Lines to append to the VCF header (see man page for details).", category: "advanced"}
+        newId: {description: "Assign ID on the fly (e.g. --set-id +'%CHROM\_%POS').", category: "advanced"}
+        include: {description: "Select sites for which the expression is true (see man page for details).", category: "advanced"}
+        keepSites: {description: "Keep sites which do not pass -i and -e expressions instead of discarding them.", category: "advanced"}
+        markSites: {description: "Annotate sites which are present ('+') or absent ('-') in the -a file with a new INFO/TAG flag.", category: "advanced"}
+        noVersion: {description: "Do not append version and command line information to the output VCF header.", category: "advanced"}
+        regions: {description: "Restrict to comma-separated list of regions.", category: "advanced"}
+        regionsFile: {description: "Restrict to regions listed in a file.", category: "advanced"}
+        renameChrs: {description: "rename chromosomes according to the map in file (see man page for details).", category: "advanced"}
+        samples: {description: "List of samples for sample stats, \"-\" to include all samples.", category: "advanced"}
+        samplesFile: {description: "File of samples to include.", category: "advanced"}
+        singleOverlaps: {description: "keep memory requirements low with very large annotation files.", category: "advanced"}
+        removeAnns: {description: "List of annotations to remove (see man page for details).", category: "advanced"}
+        inputFile: {description: "A vcf or bcf file.", category: "required"}
+
+        threads: {description: "Number of extra decompression threads [0].", category: "advanced"}
+        dockerImage: {description: "The docker image used for this task. Changing this may result in errors which the developers may choose not to address.", category: "advanced"}
+        memory: {description: "The amount of memory this job will use.", category: "advanced"}
+        timeMinutes: {description: "The maximum amount of time the job will run in minutes.", category: "advanced"}
+    }
+}
+
+task Sort {
+    input {
+        File inputFile
+        String outputPath = "output.vcf.gz"
+        String memory = "256M"
+        Int timeMinutes = 1 + ceil(size(inputFile, "G"))
+        String dockerImage = "quay.io/biocontainers/bcftools:1.10.2--h4f4756c_2"
+        String outputType = "z"
+    }
+
+    command {
+        set -e
+        mkdir -p "$(dirname ~{outputPath})"
+        bcftools sort \
+        -o ~{outputPath} \
+        -O ~{outputType} \
+        ~{inputFile}
+        bcftools index --tbi ~{outputPath}
+    }
+
+    output {
+        File outputVcf = outputPath
+        File outputVcfIndex = outputPath + ".tbi"
+    }
+    
+    runtime {
+        memory: memory
+        time_minutes: timeMinutes
+        docker: dockerImage
+    }
+
+    parameter_meta {
+        inputFile: {description: "A vcf or bcf file.", category: "required"}
+        outputPath: {description: "The location the output VCF file should be written.", category: "common"}
+        outputType: {description: "Output type: v=vcf, z=vcf.gz, b=bcf, u=uncompressed bcf", category: "advanced"}
+        memory: {description: "The amount of memory this job will use.", category: "advanced"}
+        timeMinutes: {description: "The maximum amount of time the job will run in minutes.", category: "advanced"}
+        dockerImage: {description: "The docker image used for this task. Changing this may result in errors which the developers may choose not to address.", category: "advanced"}
+    }
+
+    
+}
+
 task View {
     input {
         File inputFile

delly.wdl

@@ -28,7 +28,7 @@ task CallSV {
         File bamIndex
         File referenceFasta
         File referenceFastaFai
-        String outputPath = "./delly/delly.vcf"
+        String outputPath = "./delly/delly.bcf"
 
         String memory = "15G"
         Int timeMinutes = 300