fix: remove hard coded contig names from gcnv wrappers (#489 ) (#490)

snappy_wrappers/wrappers/gcnv/contig_ploidy/wrapper.py

@@ -43,14 +43,24 @@
 export OMP_NUM_THREADS={snakemake.threads}
 export THEANO_FLAGS="base_compiledir=$TMPDIR/theano_compile_dir"
 
+# Get contig name style
+egrep "^@SQ\s+SN:chr[0-9XY]{1,2}\s+" {snakemake.input.interval_list} > /dev/null && true
+exit_status=$?
+if [[ $exit_status == 0 ]]; then
+    STYLE="chr"
+else
+    STYLE=""
+fi
+
+
 PRIORS=$TMPDIR/ploidy_priors.tsv
 echo -e "CONTIG_NAME\tPLOIDY_PRIOR_0\tPLOIDY_PRIOR_1\tPLOIDY_PRIOR_2\tPLOIDY_PRIOR_3" \
 > $PRIORS
 for i in {{1..22}}; do
-    echo -e "$i\t0\t0.01\t0.98\t0.01" >> $PRIORS
+    echo -e "$STYLE$i\t0\t0.01\t0.98\t0.01" >> $PRIORS
 done
-echo -e "X\t0.01\t0.49\t0.49\t0.01" >> $PRIORS
-echo -e "Y\t0.495\t0.495\t0.01\t0" >> $PRIORS
+echo -e "${{STYLE}}X\t0.01\t0.49\t0.49\t0.01" >> $PRIORS
+echo -e "${{STYLE}}Y\t0.495\t0.495\t0.01\t0" >> $PRIORS
 
 set -x
 

snappy_wrappers/wrappers/gcnv/post_germline_calls/wrapper.py

@@ -25,14 +25,25 @@
     fi
 done
 
+# Get contig name style
+# Note: the sample_index in '{snakemake.input.ploidy}' and '{snakemake.input.calls[0]}' are NOT guaranteed to be the same
+# -> Sample_0 should always exist and the contig names (which we care for) are the same
+tail -n +2 $(dirname {snakemake.input.ploidy})/ploidy-calls/SAMPLE_0/contig_ploidy.tsv | egrep "^chr[0-9XY]{1,2}\s[0-9]\s[0-9.]+" > /dev/null && true
+exit_status=$?
+if [[ $exit_status == 0 ]]; then
+    STYLE="chr"
+else
+    STYLE=""
+fi
+
 gatk PostprocessGermlineCNVCalls \
     $(for x in {paths_calls}; do echo --calls-shard-path $(dirname $x)/cnv_calls-calls; done) \
     $(for x in {paths_models}; do echo --model-shard-path $(dirname $x)/cnv_calls-model; done) \
     --contig-ploidy-calls $(dirname {snakemake.input.ploidy})/ploidy-calls \
     --sample-index $sample_index \
     --autosomal-ref-copy-number 2 \
-    --allosomal-contig X \
-    --allosomal-contig Y \
+    --allosomal-contig ${{STYLE}}X \
+    --allosomal-contig ${{STYLE}}Y \
     --output-genotyped-intervals ${{itv_vcf%.gz}} \
     --output-genotyped-segments ${{seg_vcf%.gz}} \
     --output-denoised-copy-ratios {snakemake.output.ratio_tsv}

snappy_wrappers/wrappers/gcnv/post_germline_calls_case_mode/wrapper.py

@@ -29,14 +29,23 @@
     exit 1
 fi
 
+# Get contig name style; SAMPLE_0  is guaranteed to exist
+tail -n +2 $(dirname {snakemake.input.ploidy})/ploidy-calls/SAMPLE_0/contig_ploidy.tsv | grep "^chr[0-9XY]{1,2}\s[0-9]\s[0-9.]+$" > /dev/null && true
+exit_status=$?
+if [[ $exit_status == 0 ]]; then
+    STYLE="chr"
+else
+    STYLE=""
+fi
+
 gatk PostprocessGermlineCNVCalls \
     $(for x in {paths_calls}; do echo --calls-shard-path $(dirname $x)/cnv_calls-calls; done) \
     $(for x in {paths_models}; do echo --model-shard-path $x; done) \
     --contig-ploidy-calls $(dirname {snakemake.input.ploidy})/ploidy-calls \
     --sample-index $sample_index \
     --autosomal-ref-copy-number 2 \
-    --allosomal-contig X \
-    --allosomal-contig Y \
+    --allosomal-contig ${{STYLE}}X \
+    --allosomal-contig ${{STYLE}}Y \
     --output-genotyped-intervals ${{itv_vcf%.gz}} \
     --output-genotyped-segments ${{seg_vcf%.gz}} \
     --output-denoised-copy-ratios {snakemake.output.ratio_tsv}