Bcftools view (#383)

* use snakemake-wrapper-utils, allow more inputs

* update bcftools version

* update description

* add wrapper-utils to environment

* update test

* snakefmt test

* update tests

* add parsing of extra params

* add files necessary for subsetting test

* make environment compliant

* remove conda-forge dependency

* add comments to Snakefile

* readd conda-forge

bio/bcftools/view/environment.yaml

@@ -2,4 +2,5 @@ channels:
   - bioconda
   - conda-forge
 dependencies:
-  - bcftools ==1.11
+  - bcftools ==1.12
+  - snakemake-wrapper-utils ==0.2

bio/bcftools/view/meta.yaml

@@ -1,4 +1,14 @@
 name: bcftools view
-description: View vcf/bcf file in a different format. For more information see `BCFtools documentation <https://www.htslib.org/doc/bcftools.html#view>`_.
+description: View vcf/bcf file in a different format.
 authors:
   - Johannes Köster
+  - Nikos Tsardakas Renhuldt
+input:
+  - VCF/BCF file
+output:
+  - Filtered VCF/BCF file
+notes: |
+  * The `uncompressed_bcf` param allows to specify that a BCF output should be uncompressed (ignored otherwise).
+  * The `bcftools_use_mem` param controls whether to pass the resources.mem_mb to bcftools
+  * The `extra` param allows for additional program arguments (not `--threads, `-O/--output-type`, `-m/--max-mem`, or `-T/--temp-dir`).
+  * For more information see, https://samtools.github.io/bcftools/bcftools.html

bio/bcftools/view/test/Snakefile

@@ -1,11 +1,67 @@
-rule bcf_to_vcf:
+rule bcf_view_sample_file:
     input:
-        bcf="{prefix}.bcf"
+        "{prefix}.bcf",  # input bcf/vcf needs to be first input
+        index="{prefix}.bcf.csi",  # other inputs are optional
+        samples="samples.txt",
     output:
-        vcf="{prefix}.vcf"
+        "{prefix}.view_sample.vcf",
+    log:
+        "log/{prefix}.view_sample.vcf.log",
+    params:
+        # optional extra parameters
+        extra=lambda w, input: f"-S {input.samples}",
+    wrapper:
+        "master/bio/bcftools/view"
+
+
+rule bcf_view_o_vcf:
+    input:
+        "{prefix}.bcf",
+    output:
+        "{prefix}.view.vcf",
+    log:
+        "log/{prefix}.view.vcf.log",
+    params:
+        extra="",
+    wrapper:
+        "master/bio/bcftools/view"
+
+
+rule bcf_view_o_vcf_gz:
+    input:
+        "{prefix}.bcf",
+    output:
+        "{prefix}.view.vcf.gz",
+    log:
+        "log/{prefix}.view.vcf.gz.log",
+    params:
+        extra="",
+    wrapper:
+        "master/bio/bcftools/view"
+
+
+rule bcf_view_o_bcf:
+    input:
+        "{prefix}.bcf",
+    output:
+        "{prefix}.view.bcf",
+    log:
+        "log/{prefix}.view.bcf.log",
     params:
-        ""  # optional parameters for bcftools view (except -o)
+        extra="",
+    wrapper:
+        "master/bio/bcftools/view"
+
+
+rule bcf_view_o_uncompressed_bcf:
+    input:
+        "{prefix}.bcf",
+    output:
+        "{prefix}.view.uncompressed.bcf",
     log:
-        "logs/{prefix}.log"
+        "log/{prefix}.view.uncompressed.bcf.log",
+    params:
+        uncompressed_bcf=True,
+        extra="",
     wrapper:
         "master/bio/bcftools/view"

bio/bcftools/view/test/samples.txt

@@ -0,0 +1 @@
+A

bio/bcftools/view/wrapper.py

@@ -5,11 +5,16 @@
 
 
 from snakemake.shell import shell
+from snakemake_wrapper_utils.bcftools import get_bcftools_opts
 
+bcftools_opts = get_bcftools_opts(snakemake)
 extra = snakemake.params.get("extra", "")
 log = snakemake.log_fmt_shell(stdout=True, stderr=True)
 
 shell(
-    "bcftools view {extra} --threads {snakemake.threads} {snakemake.input} "
-    "-o {snakemake.output} {log}"
+    "bcftools view {bcftools_opts} "
+    "{extra} "
+    "{snakemake.input[0]} "
+    "-o {snakemake.output} "
+    "{log}"
 )

test.py

@@ -718,10 +718,42 @@ def test_bcftools_reheader():
 
 
 @skip_if_not_modified
-def test_bcftools_view():
+def test_bcftools_view_sample_vcf():
     run(
         "bio/bcftools/view",
-        ["snakemake", "--cores", "1", "--use-conda", "-F", "a.vcf"],
+        ["snakemake", "--cores", "1", "a.view_sample.vcf", "--use-conda", "-F"],
+    )
+
+
+@skip_if_not_modified
+def test_bcftools_view_vcf():
+    run(
+        "bio/bcftools/view",
+        ["snakemake", "--cores", "1", "a.view.vcf", "--use-conda", "-F"],
+    )
+
+
+@skip_if_not_modified
+def test_bcftools_view_vcf_gz():
+    run(
+        "bio/bcftools/view",
+        ["snakemake", "--cores", "1", "a.view.vcf.gz", "--use-conda", "-F"],
+    )
+
+
+@skip_if_not_modified
+def test_bcftools_view_bcf():
+    run(
+        "bio/bcftools/view",
+        ["snakemake", "--cores", "1", "a.view.bcf", "--use-conda", "-F"],
+    )
+
+
+@skip_if_not_modified
+def test_bcftools_view_uncompressed_bcf():
+    run(
+        "bio/bcftools/view",
+        ["snakemake", "--cores", "1", "a.view.uncompressed.bcf", "--use-conda", "-F"],
     )