Merge pull request #426 from broadinstitute/dp-cdc

lofreq task portability, misc other edits

Author: dpark01

pipes/WDL/tasks/tasks_intrahost.wdl

@@ -128,13 +128,25 @@ task lofreq {
 
     lofreq version | grep version | sed 's/.* \(.*\)/\1/g' | tee LOFREQ_VERSION
 
-    samtools faidx "~{reference_fasta}"
-    samtools index "~{aligned_bam}"
+    # make local copies because CWD is writeable but localization dir isn't always
+    cp "~{reference_fasta}" reference.fasta
+    cp "~{aligned_bam}" aligned.bam
+
+    # samtools faidx fails if fasta is empty
+    if [ $(grep -v '^>' reference.fasta | tr -d '\nNn' | wc -c) == "0" ]; then
+      touch "~{out_basename}.vcf"
+      exit 0
+    fi
+
+    # index for lofreq
+    samtools faidx reference.fasta
+    samtools index aligned.bam
 
+    # lofreq
     lofreq call \
-      -f "~{reference_fasta}" \
+      -f reference.fasta \
       -o "~{out_basename}.vcf" \
-      "~{aligned_bam}"
+      aligned.bam
   >>>
 
   output {

pipes/WDL/tasks/tasks_ncbi.wdl

@@ -846,7 +846,7 @@ task prepare_genbank {
   }
 }
 
-task package_genbank_ftp_submission {
+task package_sc2_genbank_ftp_submission {
   meta {
     description: "Prepares a zip and xml file for FTP-based NCBI Genbank submission according to instructions at https://www.ncbi.nlm.nih.gov/viewvc/v1/trunk/submit/public-docs/genbank/SARS-CoV-2/."
   }

pipes/WDL/workflows/metagenomic_denovo.wdl

@@ -188,6 +188,7 @@ workflow metagenomic_denovo {
     input:
       contigs_fasta           = assemble.contigs_fasta,
       reads_bam               = dehosted_bam,
+      sample_name             = sample_name,
       reference_genome_fasta  = reference_genome_fasta
   }
 

pipes/WDL/workflows/sarscov2_genbank.wdl

@@ -123,7 +123,7 @@ workflow sarscov2_genbank {
             defaults_yaml = author_sbt_defaults_yaml,
             j2_template   = author_sbt_j2_template
     }
-    call ncbi.package_genbank_ftp_submission as passing_package_genbank {
+    call ncbi.package_sc2_genbank_ftp_submission as passing_package_genbank {
       input:
         sequences_fasta          = passing_fasta.combined,
         source_modifier_table    = passing_source_modifiers.genbank_source_modifier_table,
@@ -159,7 +159,7 @@ workflow sarscov2_genbank {
         assembly_stats_tsv = assembly_stats_tsv,
         filter_to_ids      = weird_ids.ids_txt
     }
-    call ncbi.package_genbank_ftp_submission as weird_package_genbank {
+    call ncbi.package_sc2_genbank_ftp_submission as weird_package_genbank {
       input:
         sequences_fasta          = weird_fasta.combined,
         source_modifier_table    = weird_source_modifiers.genbank_source_modifier_table,

pipes/WDL/workflows/sarscov2_illumina_full.wdl

@@ -330,7 +330,7 @@ workflow sarscov2_illumina_full {
         sequences = submittable_filter.filtered_fasta,
         keep_list = [biosample_to_genbank.sample_ids]
     }
-    call ncbi.package_genbank_ftp_submission {
+    call ncbi.package_sc2_genbank_ftp_submission as package_genbank_ftp_submission {
       input:
         sequences_fasta          = submit_genomes.filtered_fasta,
         source_modifier_table    = biosample_to_genbank.genbank_source_modifier_table,

pipes/WDL/workflows/sarscov2_sra_to_genbank.wdl

@@ -199,7 +199,7 @@ workflow sarscov2_sra_to_genbank {
         assembly_stats_tsv = write_tsv(flatten([[['SeqID','Assembly Method','Coverage','Sequencing Technology']],select_all(assembly_cmt)])),
         filter_to_ids      = write_lines(select_all(submittable_id))
     }
-    call ncbi.package_genbank_ftp_submission {
+    call ncbi.package_sc2_genbank_ftp_submission as package_genbank_ftp_submission {
       input:
         sequences_fasta          = submit_genomes.combined,
         source_modifier_table    = biosample_to_genbank.genbank_source_modifier_table,