extra tools (#21)

* gatk4 versions, add latest version

* new gatk4 tool - mergebamalignment

* syntax fixes

* naming error

* new tool: GATK4 FastqToSam

* new tool: GATK4 FastqToSam

* fix merge error

* new tool: bamsormadup

* new tool: io_lib Scramble

* import and formatting fixes

* fix version imports

* fix bamsormadup tool args

* bamsormadup: fix docker url

* fix scramble tool args

* scramble: fix output type

* remove bool type from ToolArgument

* bump gatk4 version

* update fastaFai type, usually fai and dict are required together in workflows

* Revert "update fastaFai type, usually fai and dict are required together in workflows"

This reverts commit be8d880.

janis_bioinformatics/tools/__init__.py

@@ -4,6 +4,7 @@
 )
 from janis_bioinformatics.tools import (
     babrahambioinformatics,
+    biobambam,
     bcftools,
     bwa,
     common,
@@ -14,6 +15,7 @@
     gatk4,
     htslib,
     illumina,
+    io_lib,
     multiqc,
     papenfuss,
     pmac,

janis_bioinformatics/tools/biobambam/__init__.py

@@ -0,0 +1 @@
+from .bamsormadup.versions import *

janis_bioinformatics/tools/biobambam/bamsormadup/base.py

@@ -0,0 +1,175 @@
+from abc import ABC
+from typing import Any, Dict
+
+from janis_core import (
+    ToolInput,
+    ToolArgument,
+    WildcardSelector,
+    Int,
+    Float,
+    Boolean,
+    String,
+    ToolOutput,
+    Filename,
+    InputSelector,
+    CaptureType,
+    CpuSelector,
+    Stdout,
+    get_value_for_hints_and_ordered_resource_tuple,
+    ToolMetadata,
+)
+
+from janis_bioinformatics.data_types import Sam, FastaWithDict, FastqGzPair, Bam, File
+from janis_bioinformatics.tools.bioinformaticstoolbase import BioinformaticsTool
+
+BAMSORMADUP_MEM_TUPLE = [
+    (
+        CaptureType.key(),
+        {
+            CaptureType.TARGETED: 8,
+            CaptureType.EXOME: 12,
+            CaptureType.CHROMOSOME: 12,
+            CaptureType.THIRTYX: 16,
+            CaptureType.NINETYX: 20,
+            CaptureType.THREEHUNDREDX: 24,
+        },
+    )
+]
+
+BAMSORMADUP_CORES_TUPLE = [
+    (
+        CaptureType.key(),
+        {
+            CaptureType.TARGETED: 16,
+            CaptureType.EXOME: 20,
+            CaptureType.CHROMOSOME: 24,
+            CaptureType.THIRTYX: 30,
+            CaptureType.NINETYX: 32,
+            CaptureType.THREEHUNDREDX: 32,
+        },
+    )
+]
+
+
+class BamSorMaDupBase(BioinformaticsTool, ABC):
+    def tool(self):
+        return "bamsormadup"
+
+    def friendly_name(self):
+        return "BamSorMaDup"
+
+    def tool_provider(self):
+        return "BioBamBam"
+
+    def base_command(self):
+        return ["bamsormadup"]
+
+    def inputs(self):
+        return [
+            ToolInput("alignedReads", Bam(), position=200),
+            ToolInput("outputFilename", Filename(extension=".bam")),
+            *BamSorMaDupBase.additional_inputs,
+        ]
+
+    def arguments(self):
+        return [
+            ToolArgument(
+                "metrics.txt",
+                prefix="M=",
+                separate_value_from_prefix=False,
+                doc="file containing metrics from duplicate removal",
+            ),
+            ToolArgument(
+                "bam",
+                prefix="inputformat=",
+                separate_value_from_prefix=False,
+                doc="input data format",
+            ),
+            ToolArgument(
+                "bam",
+                prefix="outputFormat=",
+                separate_value_from_prefix=False,
+                doc="output data format",
+            ),
+        ]
+
+    def outputs(self):
+        return [
+            ToolOutput(
+                "out", Stdout(Bam(), stdoutname=InputSelector("outputFilename"))
+            ),
+            ToolOutput("metrics", File(), glob=WildcardSelector("metrics.txt")),
+        ]
+
+    def memory(self, hints: Dict[str, Any]):
+        val = get_value_for_hints_and_ordered_resource_tuple(
+            hints, BAMSORMADUP_MEM_TUPLE
+        )
+        if val:
+            return val
+        return 16
+
+    def cpus(self, hints: Dict[str, Any]):
+        val = get_value_for_hints_and_ordered_resource_tuple(
+            hints, BAMSORMADUP_CORES_TUPLE
+        )
+        if val:
+            return val
+        return 4
+
+    def bind_metadata(self):
+        from datetime import date
+
+        return ToolMetadata(
+            contributors=["Matthias De Smet (@mattdsm)"],
+            dateCreated=date(2020, 2, 26),
+            dateUpdated=date(2020, 2, 26),
+            institution="None",
+            doi=None,
+            keywords=["duplicates", "sort"],
+            documentationUrl="https://gitlab.com/german.tischler/biobambam2",
+            documentation="bamsormadup: parallel sorting and duplicate marking",
+        )
+
+    additional_inputs = [
+        ToolInput(
+            "level",
+            Int(optional=True),
+            prefix="level=",
+            separate_value_from_prefix=False,
+            default=0,
+            doc="compression settings for output bam file (-1=zlib default,0=uncompressed,1=fast,9=best)",
+        ),
+        ToolInput(
+            "tempLevel",
+            Int(optional=True),
+            prefix="templevel=",
+            separate_value_from_prefix=False,
+            default=0,
+            doc="compression settings for temporary bam files (-1=zlib default,0=uncompressed,1=fast,9=best)",
+        ),
+        ToolInput(
+            "threads",
+            Int(optional=True),
+            default=CpuSelector(),
+            prefix="threads=",
+            separate_value_from_prefix=False,
+            doc="Number of threads. (default = 1)",
+        ),
+        ToolInput(
+            "sortOrder",
+            String(optional=True),
+            prefix="SO=",
+            separate_value_from_prefix=False,
+            default="coordinate",
+            doc="output sort order(coordinate by default)",
+        ),
+        ToolInput(
+            "optMinPixelDif",
+            Int(optional=True),
+            prefix="optminpixeldif=",
+            separate_value_from_prefix=False,
+            default=2500,
+            doc="pixel difference threshold for optical duplicates (patterned flowcell: 12000, unpatterned flowcell: 2500)",
+        ),
+    ]

janis_bioinformatics/tools/biobambam/bamsormadup/versions.py

@@ -0,0 +1,13 @@
+from ..versions import BioBamBam_2_0_87
+from .base import BamSorMaDupBase
+
+
+class BamSorMaDup_2_0_87(BioBamBam_2_0_87, BamSorMaDupBase):
+    pass
+
+
+BamSorMaDupLatest = BamSorMaDup_2_0_87
+
+
+if __name__ == "__main__":
+    print(BamSorMaDupLatest().help())

janis_bioinformatics/tools/biobambam/versions.py

@@ -0,0 +1,12 @@
+from abc import ABC
+
+
+class BioBamBam_2_0_87(ABC):
+    def container(self):
+        return "quay.io/biocontainers/biobambam:2.0.87--1"
+
+    def version(self):
+        return "2.0.87"
+
+
+BioBamBamLatest = BioBamBam_2_0_87

janis_bioinformatics/tools/gatk4/__init__.py

@@ -18,12 +18,26 @@
     Gatk4CalculateContamination_4_1_4,
     Gatk4CalculateContaminationLatest,
 )
+from .createsequencedictionary.versions import (
+    Gatk4CreateSequenceDictionary_4_1_2,
+    Gatk4CreateSequenceDictionary_4_1_3,
+    Gatk4CreateSequenceDictionary_4_1_4,
+    Gatk4CreateSequenceDictionaryLatest,
+)
+from .fastqtosam.versions import *
 from .filtermutectcalls.versions import (
     Gatk4FilterMutectCalls_4_1_2,
     Gatk4FilterMutectCalls_4_1_3,
     Gatk4FilterMutectCalls_4_1_4,
     Gatk4FilterMutectCallsLatest,
 )
+from .gathervcfs.versions import (
+    Gatk4GatherVcfs_4_0,
+    Gatk4GatherVcfs_4_1_2,
+    Gatk4GatherVcfs_4_1_3,
+    Gatk4GatherVcfs_4_1_4,
+    Gatk4GatherVcfsLatest,
+)
 from .genotypeconcordance.versions import (
     Gatk4GenotypeConcordance_4_0,
     Gatk4GenotypeConcordance_4_1_2,
@@ -57,6 +71,7 @@
     Gatk4MarkDuplicates_4_1_4,
     Gatk4MarkDuplicatesLatest,
 )
+from .mergebamalignment.versions import *
 from .mergemutectstats.versions import (
     Gatk4MergeMutectStats_4_1_2,
     Gatk4MergeMutectStats_4_1_3,
@@ -91,24 +106,9 @@
     Gatk4SortSam_4_1_4,
     Gatk4SortSamLatest,
 )
-from .gathervcfs.versions import (
-    Gatk4GatherVcfs_4_0,
-    Gatk4GatherVcfs_4_1_2,
-    Gatk4GatherVcfs_4_1_3,
-    Gatk4GatherVcfs_4_1_4,
-    Gatk4GatherVcfsLatest,
-)
-
 from .splitreads.versions import (
+    Gatk4SortSamLatest,
     Gatk4SplitReads_4_1_2,
     Gatk4SplitReads_4_1_3,
     Gatk4SplitReads_4_1_4,
-    Gatk4SortSamLatest,
-)
-
-from .createsequencedictionary.versions import (
-    Gatk4CreateSequenceDictionary_4_1_2,
-    Gatk4CreateSequenceDictionary_4_1_3,
-    Gatk4CreateSequenceDictionary_4_1_4,
-    Gatk4CreateSequenceDictionaryLatest,
 )