fixes

docker/truvari_intrasample/Dockerfile

@@ -102,6 +102,7 @@ RUN wget https://github.com/samtools/bcftools/releases/download/${bcftools_versi
     && bcftools --help
 
 # TRUVARI
-RUN pip3 install git+https://github.com/acenglish/truvari[email protected] \
+RUN pip3 install truvari \
     && truvari --help
 COPY ./resolve.py ${work_dir}
+COPY ./inversion_guesser.py ${work_dir}

docker/truvari_intrasample/resolve.py

@@ -11,7 +11,7 @@
 import truvari
 
 REF_CLEAN = True # Set to false if you're working with the right reference
-MAX_SV = 100_000 # Filter things smaller than this
+MAX_SV = 100_000_000 # Filter things smaller than this
 
 RC = str.maketrans("ATCG", "TAGC")
 def do_rc(s):
@@ -41,6 +41,7 @@ def resolve(entry, ref):
         entry.alts = [seq]
         entry.stop = entry.start + 1
     entry.qual = 1
+
     return entry
 
 if __name__ == '__main__':
@@ -61,7 +62,13 @@ def resolve(entry, ref):
                 n_header.contigs[ctg].remove_header()
 
     out = pysam.VariantFile("/dev/stdout", 'w', header=n_header)
+    seen = set()
     for entry in vcf:
+        key = truvari.entry_to_hash(entry)
+        if key in seen:
+            continue
+        seen.add(key)
+
         if REF_CLEAN and entry.chrom not in ref.references:
             continue
 
@@ -74,9 +81,17 @@ def resolve(entry, ref):
 
         if entry is None or set(entry.alts[0]) == {'N'}:
             continue
+        if entry.info['SVTYPE'] != 'INV':
+            entry.info['SVLEN'] = abs(len(entry.ref) - len(entry.alts[0]))
+        else:
+            entry.info['SVLEN'] = len(entry.ref)
         # No more blank genotypes
         n_gt = tuple([_ if _ is not None else 0 for _ in entry.samples[0]['GT']])
+        # Preserve phasing informatino
+        is_phased = entry.samples[0].phased
         entry.samples[0]['GT'] = n_gt
+        entry.samples[0].phased = is_phased
+
         entry.translate(n_header)
         try:
             out.write(entry)

wdl/pipelines/TruvariIntrasample.wdl

@@ -89,11 +89,13 @@ task TruvariIntrasampleImpl {
         mkdir -p preprocessed
         for in_vcf in ~{pav_vcf_gz} ~{pbsv_vcf_gz} ~{sniffles_vcf_gz}
         do
+            prename=preprocessed/pre_inv_$(basename $in_vcf)
+            python ~{docker_dir}/resolve.py ${in_vcf} $ref \
+                | bcftools norm --check-ref s --fasta-ref $ref -N -m-any \
+                | bcftools view -i "SVTYPE != 'BND'" -O z -o ${prename}
+            tabix $prename
             outname=preprocessed/$(basename $in_vcf)
-            python ~{docker_dir}/resolve.py ${in_vcf} ~{reference_fa} \
-                | bcftools norm --check-ref s --fasta-ref ~{reference_fa} -N -m-any \
-                | bcftools view -i "SVTYPE != 'BND'" -O z -o ${outname}
-            tabix $outname
+            python ~{docker_dir}/inversion_guesser.py -i $prename -o $outname
         done
 
         # Step 2 - merge
@@ -108,8 +110,8 @@ task TruvariIntrasampleImpl {
         tabix ~{sample_id}.bcftools_merged.vcf.gz
 
         # Step 3 - collapse
-        truvari collapse -i ~{sample_id}.bcftools_merged.vcf.gz -c removed.vcf.gz -k maxqual --gt --intra \
-            --pctseq 0.90 --pctsize 0.90 --refdist 500 \
+        truvari collapse -i ~{sample_id}.bcftools_merged.vcf.gz -c removed.vcf.gz \
+            --sizemin 0 --sizemax 1000000 -k maxqual --gt --intra --pctseq 0.90 --pctsize 0.90 --refdist 500 \
             | bcftools sort -O z -o ~{sample_id}.truvari_collapsed.vcf.gz
         tabix ~{sample_id}.truvari_collapsed.vcf.gz
     >>>