Merge pull request #42 from NCI-GDC/feat/tumor-only

Feat/tumor only

Author: tzuni

aliquotmaf/filters/normal_depth.py

@@ -11,14 +11,19 @@ def __init__(self, cutoff):
         super().__init__(name="NormalDepth")
         self.tags = ["ndp"]
         self.cutoff = cutoff
-        self.logger.info("Using normal depth cutoff of {0}".format(cutoff))
+        if cutoff is not None:
+            self.logger.info("Using normal depth cutoff of {0}".format(cutoff))
+        else:
+            self.logger.info("Normal depth cutoff not in use")
 
     @classmethod
     def setup(cls, cutoff):
         curr = cls(cutoff)
         return curr
 
     def filter(self, maf_record):
+        if self.cutoff is None:
+            return False
         ndp = maf_record["n_depth"].value
         return ndp is not None and ndp <= self.cutoff
 

aliquotmaf/merging/record_merger/impl/v1_0.py

@@ -184,10 +184,12 @@ def maf_from_first_element(
                 selected_caller = results[caller][0]
                 break
 
+        # iterate over columns as defined by schema
         for column in self.columns:
             if column in self.allele_columns() or column == "callers":
                 continue
 
+            # do something special with depth columns
             elif column in self.average_columns(tumor_only=tumor_only):
                 vals = []
                 for caller in callers:

aliquotmaf/subcommands/merge_aliquot/runners/gdc_1_0_0_aliquot_merged.py

@@ -65,6 +65,9 @@ def __add_arguments__(cls, parser):
             "--gatk4-mutect2-pair",
             help="Path to input protected GATK4 MuTect2 Pair MAF file",
         )
+        parser.add_argument(
+            "--gatk4-mutect2", help="Path to input protected GATK4 MuTect2 MAF file"
+        )
 
     def load_readers(self):
         """
@@ -81,6 +84,7 @@ def load_readers(self):
             "caveman",
             "sanger_pindel",
             "gatk4_mutect2_pair",
+            "gatk4_mutect2",
         ]
 
         for maf_key in maf_keys:
@@ -148,15 +152,18 @@ def do_work(self):
             peekable_iterator_class=FilteringPeekableIterator,
         )
 
-        # ndp filter
-        ndp_filter = Filters.NormalDepth.setup(self.options["min_n_depth"])
+        # Set up Normal Depth Filter for tumor-only or tumor-normal operation
+        if self.options['tumor_only']:
+            ndp_filter = Filters.NormalDepth.setup(None)
+        else:
+            ndp_filter = Filters.NormalDepth.setup(self.options["min_n_depth"])
         ndp_tag = ndp_filter.tags[0]
 
         # Counts
         processed = 0
         try:
             for record in o_iter:
-
+                # progress update
                 if processed > 0 and processed % 1000 == 0:
                     self.logger.info(
                         "Processed {0} overlapping intervals...".format(processed)

aliquotmaf/subcommands/vcf_to_aliquot/extractors/effects.py

@@ -296,14 +296,16 @@ def extract(
             effect["Exon_Number"] = effect["EXON"]
             effect["Hugo_Symbol"] = effect["SYMBOL"]
             effect["TRANSCRIPT_STRAND"] = effect["STRAND"]
-            effect["1000G_AF"] = effect["AF"]
-            effect["1000G_AFR_AF"] = effect["AFR_AF"]
-            effect["1000G_AMR_AF"] = effect["AMR_AF"]
-            effect["1000G_EAS_AF"] = effect["EAS_AF"]
-            effect["1000G_EUR_AF"] = effect["EUR_AF"]
-            effect["1000G_SAS_AF"] = effect["SAS_AF"]
-            effect["ESP_AA_AF"] = effect["AA_AF"]
-            effect["ESP_EA_AF"] = effect["EA_AF"]
+
+            # these don't exist in Tumor-only
+            effect["1000G_AF"] = effect.get("AF", None)
+            effect["1000G_AFR_AF"] = effect.get("AFR_AF", None)
+            effect["1000G_AMR_AF"] = effect.get("AMR_AF", None)
+            effect["1000G_EAS_AF"] = effect.get("EAS_AF", None)
+            effect["1000G_EUR_AF"] = effect.get("EUR_AF", None)
+            effect["1000G_SAS_AF"] = effect.get("SAS_AF", None)
+            effect["ESP_AA_AF"] = effect.get("AA_AF", None)
+            effect["ESP_EA_AF"] = effect.get("EA_AF", None)
 
             # If VEP couldn't find this variant in dbSNP/etc., we'll say it's "novel"
             if effect["Existing_variation"]: