Merge pull request #95 from dufeiyu/update

Use MSHDRepo, launcher script update and QC metrics change

Author: dufeiyu

MyeloseqHD.wdl

@@ -51,7 +51,7 @@ workflow MyeloseqHD {
     String CustomAnnotationIndex = MyeloSeqHDRepo + "/accessory_files/MyeloseqHD.custom_annotations.vcf.gz.tbi"
     String CustomAnnotationParameters = "MYELOSEQ,vcf,exact,0,TCGA_AC,MDS_AC,BLACKLIST"
 
-    String QC_pl   = "/usr/local/bin/QC_metrics.pl"
+    String QC_pl   = MyeloSeqHDRepo + "/dockerfiles/docker-myeloseq/QC_metrics.pl"
     String xfer_pl = MyeloSeqHDRepo + "/scripts/data_transfer.pl"
 
     String DemuxFastqDir = "/storage1/fs1/gtac-mgi/Active/CLE/assay/myeloseqhd/demux_fastq"

MyeloseqHDAnalysis.wdl

@@ -106,6 +106,7 @@ workflow MyeloseqHDAnalysis {
                Vaf=MinVaf,
                Reads=MinReads,
                Name=Name,
+               MyeloSeqHDRepo=MyeloSeqHDRepo,
                queue=Queue,
                jobGroup=JobGroup
     }
@@ -313,6 +314,7 @@ task clean_variants {
 task combine_variants {
     input{
         Array[String] Vcfs
+        String MyeloSeqHDRepo
         String Cram
         String CramIndex
         String refFasta
@@ -328,7 +330,7 @@ task combine_variants {
      command {
          /usr/local/bin/bcftools merge -F x -m none --force-samples -Oz ${sep=" " Vcfs} | /usr/local/bin/bcftools sort -Oz -o combined.vcf.gz && \
          /usr/bin/tabix -p vcf combined.vcf.gz && \
-         /usr/bin/python3 /home/fdu/git/cle-myeloseqhd/dockerfiles/docker-myeloseq/filterHaloplex.py -r ${refFasta} --minreadsperfamily ${default='3' MinReadsPerFamily} -m ${default='5' Reads} -d ${default='0.02' Vaf} combined.vcf.gz ${Cram} ${Name} > ${Name}.combined_and_tagged.vcf
+         /usr/bin/python3 ${MyeloSeqHDRepo}/dockerfiles/docker-myeloseq/filterHaloplex.py -r ${refFasta} --minreadsperfamily ${default='3' MinReadsPerFamily} -m ${default='5' Reads} -d ${default='0.02' Vaf} combined.vcf.gz ${Cram} ${Name} > ${Name}.combined_and_tagged.vcf
      }
 
      runtime {

accessory_files/MyeloseqHD.QCMetrics.json

@@ -20,6 +20,6 @@
     "Failed hotspot": "0",
     "Haplotect sites": ",10",
     "Haplotect score": ",0.02",
-    "ASSAY VERSION": "2.2",
+    "ASSAY VERSION": "2.3",
     "DISCLAIMER": "This laboratory developed test (LDT) was developed and its performance characteristics determined by the CLIA Licensed Environment laboratory at the McDonnell Genome Institute at Washington University (MGI-CLE, CLIA #26D2092546, CAP #9047655), Dr. David H. Spencer MD, PhD, FCAP, Medical Director. 4444 Forest Park Avenue, Rm 4111 St. Louis, Missouri 63108 (314) 286-1460 Fax: (314) 286-1810. The MGI-CLE laboratory is regulated under CLIA as certified to perform high-complexity testing. Interpretation of sequencing results and case sign out is performed by Pathology and Immunology faculty in the Division of Anatomic and Molecular Pathology (WU-AMP Core Laboratory, CLIA #26D2013203, CAP #7233522) Dr. Fouad Boulos, MD, Medical Director. Pathology Consult Services, 425 S. Euclid Ave, Suite 4711, Mailstop 8024-14-4711, St. Louis, MO 63110 (314)-362-7784 Fax: (314) 362-4080. The AMP core laboratory is regulated under CLIA as certified to perform high-complexity testing. This test has not been cleared or approved by the FDA."
 }

dockerfiles/docker-myeloseq/QC_metrics.pl

@@ -35,7 +35,7 @@
     PCT_MAPPED_READS   => 'MAPPING/ALIGNING SUMMARY: Mapped reads (%)',
     MEAN_INSERT_SIZE   => 'MAPPING/ALIGNING SUMMARY: Insert length: mean',
     PCT_TARGET_BASES_GT_250x  => 'COVERAGE SUMMARY: Target bases >250x (%)',
-    PCT_TARGET_BASES_GT_2000x => 'COVERAGE SUMMARY: Target bases >2000x (%)',
+    PCT_TARGET_BASES_GT_1250x => 'COVERAGE SUMMARY: Target bases >1250x (%)',
     PCT_TARGET_ALIGNED_READS  => 'COVERAGE SUMMARY: Aligned reads in target region (%)',
     AVG_ALIGN_TARGET_COVERAGE => 'COVERAGE SUMMARY: Average alignment coverage over target region',
     PCT_LOW_COVERAGE_AMPLICON => 'AMPLICON SUMMARY: Amplicons with low coverage (%)',

imports.zip

@@ -149,25 +149,6 @@
     $seq_id++;
 }
 
-## DRAGEN sample sheet
-my $dragen_ss  = File::Spec->join($out_dir, 'demux_sample_sheet.csv'); 
-my $ss_fh = IO::File->new(">$dragen_ss") or die "Fail to write to $dragen_ss";
-$ss_fh->print("[Settings]\n");
-$ss_fh->print("AdapterBehavior,trim\n");
-$ss_fh->print("AdapterRead1,AAGATCGGAAGAGCACACGTCTGAACTCC+CAGATCGGAAGAGCACACGTCTGAACTCC+GAGATCGGAAGAGCACACGTCTGAACTCC+TAGATCGGAAGAGCACACGTCTGAACTCC\n");
-$ss_fh->print("AdapterRead2,AAAGATCGGAAGAGCGTCGTGTAGGGAAA+CAAGATCGGAAGAGCGTCGTGTAGGGAAA+GAAGATCGGAAGAGCGTCGTGTAGGGAAA+TAAGATCGGAAGAGCGTCGTGTAGGGAAA\n");
-$ss_fh->print("OverrideCycles,N1Y150;I8N2;U10;N1Y150\n");
-$ss_fh->print("[Data]\n");
-$ss_fh->print("Lane,Sample_ID,Sample_Name,Sample_Project,index,index2\n");
-$ss_fh->print($ds_str);
-$ss_fh->close;
-
-## Sample Index
-my $si = File::Spec->join($out_dir, 'sample_index');
-my $si_fh = IO::File->new(">$si") or die "Fail to write to $si";
-$si_fh->print($si_str);
-$si_fh->close;
-
 ## Get RunInfoString
 my $run_xml = File::Spec->join($rundir, 'RunParameters.xml');
 unless (-s $run_xml) {
@@ -207,6 +188,36 @@
 }
 $xml_fh->close;
 
+## DRAGEN sample sheet
+my $num_N;
+if ($index1cycle == 19) {
+    $num_N = 11;
+}
+elsif ($index1cycle == 10) {
+    $num_N = 2;
+}
+else {
+    die "Invalid index1cycle $index1cycle";
+}
+
+my $dragen_ss  = File::Spec->join($out_dir, 'demux_sample_sheet.csv');
+my $ss_fh = IO::File->new(">$dragen_ss") or die "Fail to write to $dragen_ss";
+$ss_fh->print("[Settings]\n");
+$ss_fh->print("AdapterBehavior,trim\n");
+$ss_fh->print("AdapterRead1,AAGATCGGAAGAGCACACGTCTGAACTCC+CAGATCGGAAGAGCACACGTCTGAACTCC+GAGATCGGAAGAGCACACGTCTGAACTCC+TAGATCGGAAGAGCACACGTCTGAACTCC\n");
+$ss_fh->print("AdapterRead2,AAAGATCGGAAGAGCGTCGTGTAGGGAAA+CAAGATCGGAAGAGCGTCGTGTAGGGAAA+GAAGATCGGAAGAGCGTCGTGTAGGGAAA+TAAGATCGGAAGAGCGTCGTGTAGGGAAA\n");
+$ss_fh->print('OverrideCycles,N1Y150;I8N'.$num_N.";U10;N1Y150\n");
+$ss_fh->print("[Data]\n");
+$ss_fh->print("Lane,Sample_ID,Sample_Name,Sample_Project,index,index2\n");
+$ss_fh->print($ds_str);
+$ss_fh->close;
+
+## Sample Index
+my $si = File::Spec->join($out_dir, 'sample_index');
+my $si_fh = IO::File->new(">$si") or die "Fail to write to $si";
+$si_fh->print($si_str);
+$si_fh->close;
+
 my $run_info_str = join ',', $runid, $instr, $side, $fcmode, $wftype, $R1cycle, $index1cycle, $index2cycle, $R2cycle; 
 
 ## Input JSON