IMAGE_vqsr : Execute VQSR on split SNP/INDEL GATK variant, not on full GATK variants

Snakemake/IMAGE_vqsr/rules/split_vcf.smk

@@ -10,20 +10,28 @@ wildcard_constraints:
     var = "SNP|INDEL"
     
 def get_input_vcf(wildcards) :
+    print(wildcards.prefix)
     if wildcards.prefix == "common" :
+        print("results/intersect/common_variants.vcf.gz")
         return "results/intersect/common_variants.vcf.gz"
-    if wildcards.prefix == "VQSR_trainingSet_untrusted_uniquely_called_variants":
+    elif wildcards.prefix == "VQSR_trainingSet_untrusted_uniquely_called_variants":
+        print("results/intersect/VQSR_trainingSet_untrusted_uniquely_called_variants.vcf.gz")
         return "results/intersect/VQSR_trainingSet_untrusted_uniquely_called_variants.vcf.gz"
+    elif wildcards.prefix == gatk_prefix:
+        return config['GATK_variants']
     else:
         for Set in config['vqsr_resources']:
             if os.path.basename(Set['file']).startswith(wildcards.prefix):
                 return Set['file']
-                
+        print("problem")
+                                
 def get_input_index(wildcards) :
     if wildcards.prefix == "common" :
         return "results/intersect/common_variants.vcf.gz.tbi"
-    if wildcards.prefix == "VQSR_trainingSet_untrusted_uniquely_called_variants":
+    elif wildcards.prefix == "VQSR_trainingSet_untrusted_uniquely_called_variants":
         return "results/intersect/VQSR_trainingSet_untrusted_uniquely_called_variants.vcf.gz.tbi"
+    elif wildcards.prefix == gatk_prefix:
+        return config['GATK_variants'] + ".tbi"
     else:
         for Set in config['vqsr_resources']:
             if os.path.basename(Set['file']).startswith(wildcards.prefix):

Snakemake/IMAGE_vqsr/rules/vqsr.smk

@@ -23,7 +23,7 @@ def get_trainingSet(wildcards):
 rule VariantRecalibrator :
     input :
         ref = config["reference_genome"],
-        vcf = config["GATK_variants"],
+        vcf = "results/split_SNP_INDEL/{GATK_input_prefix}_{var}.vcf.gz",
         resources = get_trainingSet
     output :
         recalFile = temp("results/VQSR/{GATK_input_prefix}_{var}.recal"),
@@ -50,7 +50,7 @@ rule VariantRecalibrator :
 rule ApplyRecalibration:
     input:
         ref = config["reference_genome"],
-        vcf = config["GATK_variants"],
+        vcf = "results/split_SNP_INDEL/{GATK_input_prefix}_{var}.vcf.gz",
         recalFile = "results/VQSR/{GATK_input_prefix}_{var}.recal",
         tranchesFile = "results/VQSR/{GATK_input_prefix}_{var}.tranches",
         R = "results/VQSR/{GATK_input_prefix}_{var}.R"