Make pangolin update optional (#48)

* Make pangolin update optional

* Update conda env directory name

.github/scripts/create_conda_env.sh

@@ -4,4 +4,4 @@ set -eo pipefail
 
 export PATH=/opt/miniconda3/bin:$PATH
 
-mamba env create --prefix /home/runner/.conda/envs/ncov-qc-724a4524bfccdf191bf8550202dac824 --file /home/runner/work/ncov-tools-nf/ncov-tools-nf/environments/ncov-tools-1.5.yml
+mamba env create --prefix /home/runner/.conda/envs/ncov-qc-40e1fefdb1b312663435a35c6d6cb9da --file /home/runner/work/ncov-tools-nf/ncov-tools-nf/environments/ncov-tools-1.5.yml

README.md

@@ -13,20 +13,23 @@ A `metadata.tsv` file may be optionally provided. If it is provided, it should f
 
 The `completeness_threshold` can optionally be modified. Its default value is the same as that of `ncov-tools`, `0.75`.
 
+The `--update_pangolin` flag allows control over whether or not pangolin will be updated prior to running ncov-tools.
+
 In the command below, `[square brackets]` indicate optional parameters. `<angled brackets>` indicate a placeholder to be replaced by the user in a real pipeline invocation.
 
 ```
 nextflow run BCCDC-PHL/ncov-tools-nf \
   -profile conda \
   --cache ~/.conda/envs \
-  --run_name <run_name> \
-  --artic_analysis_dir <output/from/ncov2019-artic-nf> \
+  [--update_pangolin] \
   [--completeness_threshold <completeness_threshold>] \
   [--metadata <metadata.tsv>] \
   [--downsampled] \
   [--split_by_plate] \
   [--freebayes_consensus] \
   [--freebayes_variants] \
+  --run_name <run_name> \
+  --artic_analysis_dir <output/from/ncov2019-artic-nf> \
   --outdir <ncov-tools-output> \
 ```
 

environments/ncov-tools-1.5.yml

@@ -23,13 +23,16 @@ dependencies:
   - poppler
   - gofasta
   - numpy=1.19.1
-  - biopython=1.77
+  - biopython=1.74
   - snpeff=5.0-0
+  - pangolin=2.4.2
+  - pangolearn
   - pip:
       - dendropy>=4.4.0
       - pyvcf
       - ncov-parser
       - git+https://github.com/hCoV-2019/lineages.git
-      - git+https://github.com/hCoV-2019/pangolin.git
-      - git+https://github.com/cov-lineages/pangoLEARN.git
       - git+https://github.com/jts/ncov-watch.git
+      - git+https://github.com/cov-lineages/scorpio.git
+      - git+https://github.com/cov-lineages/constellations.git
+

main.nf

@@ -35,7 +35,7 @@ workflow {
   ch_artic_analysis_dir = Channel.fromPath(params.artic_analysis_dir, type: 'dir')
   ch_metadata = Channel.fromPath(params.metadata, type: 'file')
 
-  update_pangolin(Channel.value(true))
+  update_pangolin(Channel.value(params.update_pangolin))
   download_ncov_tools(ch_ncov_tools_version)
   download_artic_ncov2019(ch_primer_scheme_version.combine(ch_primer_scheme_name))
   download_ncov_watchlists(ch_ncov_watchlists_version)

modules/ncov-tools.nf

@@ -1,15 +1,24 @@
 process update_pangolin {
+
+  tag { should_update.toString() }
+
   executor 'local'
 
   input:
   val(should_update)
 
   output:
-  val(true)
+  val(did_update)
 
   script:
-  """
-  pangolin --update
+  did_update = should_update
+  should_update_string = should_update ? "true" : "false"
+  """
+  should_update=${should_update_string}
+  if [ "\$should_update" = true ]
+  then
+    pangolin --update
+  fi
   """
 }
 

nextflow.config

@@ -9,6 +9,7 @@ params {
   split_by_plate = false
   freebayes_consensus = false
   freebayes_variants = false
+  update_pangolin = false
 }
 
 profiles {
@@ -29,5 +30,5 @@ process {
 manifest {
   mainScript = 'main.nf'
   nextflowVersion = '>=20.01.0'
-  version = '1.5.4'
+  version = '1.5.5'
 }