feat: make it possible to set output type, vcf or gvcf. (#476)

* feat: make it possible to set output type, vcf or gvcf.

* fix: lock mamba version.

* Update wrapper.py

* Update main.yml

* Update main.yml

* Update qc.yml

* Update qc.yml

bio/gatk/haplotypecaller/test/Snakefile

@@ -1,4 +1,24 @@
 rule haplotype_caller:
+    input:
+        # single or list of bam files
+        bam="mapped/{sample}.bam",
+        ref="genome.fasta",
+        # known="dbsnp.vcf"  # optional
+    output:
+        vcf="calls/{sample}.vcf",
+    # 	bam="{sample}.assemb_haplo.bam",
+    log:
+        "logs/gatk/haplotypecaller/{sample}.log",
+    params:
+        extra="",  # optional
+        java_opts="",  # optional
+    threads: 4
+    resources:
+        mem_mb=1024,
+    wrapper:
+        "master/bio/gatk/haplotypecaller"
+
+rule haplotype_caller_gvcf:
     input:
         # single or list of bam files
         bam="mapped/{sample}.bam",

bio/gatk/haplotypecaller/wrapper.py

@@ -27,9 +27,25 @@
 if known:
     known = "--dbsnp " + str(known)
 
+vcf_output = snakemake.output.get("vcf", "")
+if vcf_output:
+    output = " --output " + str(vcf_output)
+
+gvcf_output = snakemake.output.get("gvcf", "")
+if gvcf_output:
+    output = " --emit-ref-confidence GVCF " + " --output " + str(gvcf_output)
+
+if (vcf_output and gvcf_output) or (not gvcf_output and not vcf_output):
+    if vcf_output and gvcf_output:
+        raise ValueError(
+            "please set vcf or gvcf as output, not both! It's not supported by gatk"
+        )
+    else:
+        raise ValueError("please set one of vcf or gvcf as output (not both)!")
+
 bam_output = snakemake.output.get("bam", "")
 if bam_output:
-    bam_output = "--bam-output " + str(bam_output)
+    bam_output = " --bam-output " + str(bam_output)
 
 log = snakemake.log_fmt_shell(stdout=True, stderr=True)
 
@@ -43,7 +59,7 @@
         " {known}"
         " {extra}"
         " --tmp-dir {tmpdir}"
-        " --emit-ref-confidence GVCF {bam_output}"
-        " --output {snakemake.output.gvcf}"
+        " {output}"
+        " {bam_output}"
         " {log}"
     )

test.py

@@ -3160,7 +3160,14 @@ def test_gatk_applybqsrspark():
 
 
 @skip_if_not_modified
-def test_gatk_haplotypecaller():
+def test_gatk_haplotypecaller_vcf():
+    run(
+        "bio/gatk/haplotypecaller",
+        ["snakemake", "--cores", "1", "calls/a.vcf", "--use-conda", "-F"],
+    )
+
+@skip_if_not_modified
+def test_gatk_haplotypecaller_gvcf():
     run(
         "bio/gatk/haplotypecaller",
         ["snakemake", "--cores", "1", "calls/a.g.vcf", "--use-conda", "-F"],