tumor-only prioritization: do not apply LowPriority filter by default

Annotating input files with external (ExAC, 1000 genomes, ClinVar, COSMIC)
membership information, but do not filter by default. Filters like
impact were too stringent for users, so this provides a more flexible input
to avoid over filtering. The goal is to move the determination of germline
status to PureCN using DB and POP_AF flags.

HISTORY.md

@@ -1,5 +1,8 @@
 ## 1.1.1 (in progress)
 
+- tumor-only prioritization: do not apply LowPriority filter by default, instead
+  annotate with external databases. Use `tumoronly_germline_filter` to re-enable
+  previous behavior.
 - gemini: databases no longer created by default. Use `tools_on: [gemini]` or
   `tools_on: [gemini_orig]` to create a database.
 - vcfanno: run gemini and somatic annotations by default, producing annotated

bcbio/variation/prioritize.py

@@ -15,11 +15,11 @@
 import csv
 import re
 
-from bcbio import install, utils
+from bcbio import utils
 from bcbio.distributed.transaction import file_transaction
 from bcbio.pipeline import datadict as dd
 from bcbio.provenance import do
-from bcbio.variation import population, vcfanno, vcfutils
+from bcbio.variation import population, vcfutils
 
 geneimpacts = utils.LazyImport("geneimpacts")
 cyvcf2 = utils.LazyImport("cyvcf2")
@@ -33,7 +33,7 @@ def handle_vcf_calls(vcf_file, data, orig_items):
         ann_vcf = population.run_vcfanno(vcf_file, data)
         if ann_vcf:
             priority_file = _prep_priority_filter_vcfanno(ann_vcf, data)
-            return _apply_priority_filter(vcf_file, priority_file, data)
+            return _apply_priority_filter(ann_vcf, priority_file, data)
         # No data available for filtering, return original file
         else:
             return vcf_file
@@ -45,14 +45,19 @@ def _apply_priority_filter(in_file, priority_file, data):
     if not utils.file_exists(out_file):
         with file_transaction(data, out_file) as tx_out_file:
             header = ('##INFO=<ID=EPR,Number=.,Type=String,'
-                      'Description="Prioritization based on external annotations">')
+                      'Description="Somatic prioritization based on external annotations, '
+                      'identify as likely germline">')
             header_file = "%s-repeatheader.txt" % utils.splitext_plus(tx_out_file)[0]
             with open(header_file, "w") as out_handle:
                 out_handle.write(header)
+            if "tumoronly_germline_filter" in dd.get_tools_on(data):
+                filter_cmd = ("bcftools filter -m '+' -s 'LowPriority' "
+                              """-e "EPR[*] != 'pass'" |""")
+            else:
+                filter_cmd = ""
             cmd = ("bcftools annotate -a {priority_file} -h {header_file} "
                    "-c CHROM,FROM,TO,REF,ALT,INFO/EPR {in_file} | "
-                   "bcftools filter -m '+' -s 'LowPriority' "
-                   """-e "EPR[*] != 'pass'" | bgzip -c > {tx_out_file}""")
+                   "{filter_cmd} bgzip -c > {tx_out_file}")
             do.run(cmd.format(**locals()), "Run external annotation based prioritization filtering")
     vcfutils.bgzip_and_index(out_file, data["config"])
     return out_file
@@ -145,12 +150,10 @@ def _calc_priority_filter(row, pops):
 
     - Pass high/medium impact variants not found in population databases
     - Pass variants found in COSMIC or Clinvar provided they don't have two
-      additional reasons to filter (low severity or found in multiple external populations)
+      additional reasons to filter (found in multiple external populations)
     """
     filters = []
     passes = []
-    if row.get("impact_severity") in ["LOW"]:
-        filters.append("lowseverity")
     passes.extend(_find_known(row))
     filters.extend(_known_populations(row, pops))
     if len(filters) == 0 or (len(passes) > 0 and len(filters) < 2):

docs/contents/configuration.rst

@@ -1083,6 +1083,10 @@ lists with multiple options:
   - ``qualimap_full`` runs Qualimap with full bam files but it may be slow.
   - ``damage_filter`` annotates low frequency somatic calls in INFO/DKFZBias for
     DNA damage artifacts using `DKFZBiasFilter <https://github.com/eilslabs/DKFZBiasFilter>`_.
+  - ``tumoronly_germline_filter`` applies a ``LowPriority`` filter to tumor-only calls
+    that match population germline databases. The default is to just apply a tag
+    ``EPR`` (external prioritization) that flags variants present in external
+    databases. Anything missing a ``pass`` here is a likely germline.
   - ``vqsr`` makes GATK try quality score recalibration for variant filtration,
     even for smaller sample sizes.
   - ``svplots`` adds additional coverage and summary plots for CNVkit and detected