Merge pull request #199 from theiagen/smw-tbprofiler-continuing-dev

Smw tbprofiler continuing dev
theiagen · Sep 29, 2023 · 34cec39 · 34cec39
2 parents f008d92 + c3a5849
commit 34cec39
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 7 deletions.
diff --git a/tasks/species_typing/task_tbp_parser.wdl b/tasks/species_typing/task_tbp_parser.wdl
@@ -13,7 +13,7 @@ task tbp_parser {
     Int coverage_threshold = 100
     Boolean tbp_parser_debug = false
 
-    String docker = "us-docker.pkg.dev/general-theiagen/theiagen/tbp-parser:0.0.10"
+    String docker = "us-docker.pkg.dev/general-theiagen/theiagen/tbp-parser:1.0.0"
     Int disk_size = 100
     Int memory = 4
     Int cpu = 1

diff --git a/tasks/species_typing/task_tbprofiler.wdl b/tasks/species_typing/task_tbprofiler.wdl
@@ -16,6 +16,8 @@ task tbprofiler {
     Int cov_frac_threshold = 1
     Int cpu = 8 
     Boolean ont_data = false
+    File? tbprofiler_custom_db
+    Boolean tbprofiler_run_custom_db = false
   }
   command <<<
     # Print and save date
@@ -38,6 +40,26 @@ task tbprofiler {
       export ont_data="false"
     fi
 
+    # check if new database file is provided and not empty
+    if [ "~{tbprofiler_run_custom_db}" = true ] ; then
+
+      echo "Found new database file ~{tbprofiler_custom_db}"
+      prefix=$(basename "~{tbprofiler_custom_db}" | sed 's/\.tar\.gz$//')
+      echo "New database will be created with prefix $prefix"
+
+      echo "Inflating the new database..."
+      tar xfv ~{tbprofiler_custom_db}
+
+      tb-profiler load_library ./"$prefix"/"$prefix"
+
+      TBDB="--db $prefix"
+
+    else
+
+      TBDB=""
+
+    fi
+
     # Run tb-profiler on the input reads with samplename prefix
     tb-profiler profile \
       ${mode} \
@@ -50,7 +72,8 @@ task tbprofiler {
       --reporting_af \
       ~{min_af_pred} \
       --coverage_fraction_threshold ~{cov_frac_threshold} \
-      --csv --txt
+      --csv --txt \
+      $TBDB
 
     # Collate results
     tb-profiler collate --prefix ~{samplename}

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -620,7 +620,7 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_sonneityping.wdl
       md5sum: aeb12cf9a0db4e34f2aecbaba51c30fb
     - path: miniwdl_run/wdl/tasks/species_typing/task_tbprofiler.wdl
-      md5sum: f3c9cdca6d49878ab2be31aff128e1b4
+      md5sum: c2a8c0978cc1bd65656584e1bad9dca7
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: d49ae0b02e798af0636eb2721bb434b4
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
@@ -636,7 +636,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
       md5sum: ca8825e5cc5a6d910e0368fcb6992905
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: c4861a59d49b13b67706631a0e1246c4
+      md5sum: 2614fac622fa2035b80a7b220b1aed86
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
       contains: ["version", "QC", "output"]
     - path: miniwdl_run/workflow.log

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -588,7 +588,7 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_sonneityping.wdl
       md5sum: aeb12cf9a0db4e34f2aecbaba51c30fb
     - path: miniwdl_run/wdl/tasks/species_typing/task_tbprofiler.wdl
-      md5sum: f3c9cdca6d49878ab2be31aff128e1b4
+      md5sum: c2a8c0978cc1bd65656584e1bad9dca7
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: d49ae0b02e798af0636eb2721bb434b4
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
@@ -604,7 +604,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
       md5sum: 627a157d962453229f619da37e03e43d
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: c4861a59d49b13b67706631a0e1246c4
+      md5sum: 2614fac622fa2035b80a7b220b1aed86
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl
       md5sum: 53d322d895837c0bcb049786572e944d
     - path: miniwdl_run/workflow.log

diff --git a/workflows/utilities/wf_merlin_magic.wdl b/workflows/utilities/wf_merlin_magic.wdl
@@ -62,6 +62,8 @@ workflow merlin_magic {
     Boolean call_shigeifinder_reads_input = false
     Boolean assembly_only = false
     Boolean theiaeuk = false
+    Boolean tbprofiler_run_custom_db = false
+    File tbprofiler_custom_db = "gs://theiagen-public-files/terra/theiaprok-files/tbdb_varpipe_combined_nodups.tar.gz"
     Boolean tbprofiler_additional_outputs = false
     String tbp_parser_output_seq_method_type = "WGS"
     String? tbp_parser_operator
@@ -247,7 +249,9 @@ workflow merlin_magic {
         input:
           read1 = select_first([clockwork_decon_reads.clockwork_cleaned_read1, read1]),
           read2 = select_first([clockwork_decon_reads.clockwork_cleaned_read2, read2, "gs://theiagen-public-files/terra/theiaprok-files/no-read2.txt"]),
-          samplename = samplename
+          samplename = samplename,
+          tbprofiler_run_custom_db = tbprofiler_run_custom_db,
+          tbprofiler_custom_db = tbprofiler_custom_db
       }
       if (tbprofiler_additional_outputs) {
         call tbp_parser_task.tbp_parser {