Include profile for open 21L builds

.github/workflows/rebuild-open-21L.yml

@@ -0,0 +1,90 @@
+name: Rebuild open 21L phylogenetic datasets
+
+on:
+  # This workflow can be triggered from repository_dispatch events,
+  # for instance, after the appropriate preprocessing actions have completed
+  repository_dispatch:
+    types:
+      - rebuild
+      - open/rebuild
+      - genbank/rebuild
+  # Manually triggered using GitHub's UI
+  workflow_dispatch:
+    inputs:
+      trial_name:
+        description: "Short name for this trial build, for prefixing the uploaded data and results files. WARNING: without this we will overwrite files in s3://nextstrain-data/files/ncov/open/ and the trees on nextstrain.org/ncov/open/..."
+        required: false
+      image:
+        description: 'Specific container image to use for build (will override the default of "nextstrain build")'
+        required: false
+
+env:
+  TRIAL_NAME: ${{ github.event.inputs.trial_name }}
+  NEXTSTRAIN_DOCKER_IMAGE: ${{ github.event.inputs.image }}
+
+jobs:
+  open-21L:
+    runs-on: ubuntu-22.04
+    steps:
+    - uses: actions/checkout@v4
+
+    - uses: nextstrain/.github/actions/setup-nextstrain-cli@master
+
+    - name: Launch build
+      run: |
+        set -x
+
+        declare -a config
+        config+=(build_date=\'$(date +'%Y-%m-%d')\')
+        if [[ "$TRIAL_NAME" ]]; then
+          config+=(
+            S3_DST_BUCKET=nextstrain-staging/files/ncov/open/trial/"$TRIAL_NAME"
+            deploy_url=s3://nextstrain-staging/
+            auspice_json_prefix=ncov_open_21L_trial_"$TRIAL_NAME"
+          )
+        else
+          config+=(slack_token=$SLACK_TOKEN)
+        fi
+
+        nextstrain build \
+          --aws-batch \
+          --detach \
+          --cpus 72 \
+          --memory 140GiB \
+          . \
+            deploy \
+            upload \
+            --config "${config[@]}" \
+            --profile nextstrain_profiles/nextstrain-open-21L \
+            --set-threads tree=8 \
+        |& tee build-launch.log
+      env:
+        AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+        AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+        SLACK_TOKEN: ${{ secrets.SLACK_TOKEN }}
+
+    - name: Build info
+      run: |
+        if [[ "$TRIAL_NAME" ]]; then
+          echo "--> Trial name is: $TRIAL_NAME"
+          echo
+          echo "--> When completed, the following will be available:"
+          echo "build files: s3://nextstrain-staging/files/ncov/open/trial/$TRIAL_NAME/"
+          echo "nextstrain URLs: https://nextstrain.org/staging/ncov/open/21L/trial/$TRIAL_NAME/REGION_NAME/TIME_SPAN"
+          echo "e.g. https://nextstrain.org/staging/ncov/open/21L/trial/$TRIAL_NAME/global/all-time"
+        else
+          echo "--> open 21L phylogenetic analysis rebuilding on AWS"
+          echo
+          echo "--> When completed, the following will be updated:"
+          echo "build files: s3://nextstrain-data/files/ncov/open/REGION_NAME"
+          echo "nextstrain URLs: https://nextstrain.org/ncov/open/21L/REGION_NAME/TIME_SPAN"
+          echo "e.g. https://nextstrain.org/ncov/open/21L/global/all-time"
+        fi
+        echo
+        echo "--> You can attach to this AWS job via:"
+        tail -n1 build-launch.log
+        echo
+        JOBID=$( tail -n1 build-launch.log | sed -E 's/.+attach ([-a-f0-9]+).+/\1/' )
+        echo "--> View this job in the AWS console via"
+        echo "    https://console.aws.amazon.com/batch/home?region=us-east-1#jobs/detail/${JOBID}"
+        echo

docs/src/reference/change_log.md

@@ -5,6 +5,8 @@ We also use this change log to document new features that maintain backward comp
 
 ## New features since last version update
 
+- 9 March 2024: Include 21L-focused builds for open data, eg https://nextstrain.org/ncov/open/21L/global/6m. [PR 1101](https://github.com/nextstrain/ncov/pull/1101)
+
 - 31 January 2024: Remove RBD-level related rules and files since this feature has been broken since May 2023 and is no longer relevant. [PR 1097](https://github.com/nextstrain/ncov/pull/1097)
 
 - 30 January 2024: Fix RBD-level coloring by updating clade label and clade parsing. [PR 1094](https://github.com/nextstrain/ncov/pull/1094)

nextstrain_profiles/nextstrain-gisaid-21L/builds.yaml

@@ -889,6 +889,9 @@ refine:
 # if different traits should be reconstructed for some builds, specify here
 # otherwise the default trait config in defaults/parameters.yaml will used
 traits:
+  reference:
+    sampling_bias_correction: 2.5
+    columns: ["region"]
   global_1m:
     sampling_bias_correction: 2.5
     columns: ["region"]
@@ -979,6 +982,10 @@ traits:
 # narrow_bandwidth = 0.019 or 7 days for "1m" and "2m"
 # narrow_bandwidth = 0.038 or 14 days for "6m" and "all-time"
 frequencies:
+  reference:
+    min_date: "2022-01-01"
+    narrow_bandwidth: 0.038
+    recent_days_to_censor: 7
   global_1m:
     min_date: "1M"
     narrow_bandwidth: 0.019

nextstrain_profiles/nextstrain-gisaid-21L/exclude-clades.tsv

@@ -5,21 +5,21 @@ clade
 20B
 20C
 20D
-20E (EU1)
+20E
 20F
 20G
-20H (Beta, V2)
-20I (Alpha, V1)
-20J (Gamma, V3)
-21A (Delta)
-21B (Kappa)
-21C (Epsilon)
-21D (Eta)
-21E (Theta)
-21F (Iota)
-21G (Lambda)
-21H (Mu)
-21I (Delta)
-21J (Delta)
-21K (Omicron)
-21M (Omicron)
+20H
+20I
+20J
+21A
+21B
+21C
+21D
+21E
+21F
+21G
+21H
+21I
+21J
+21K
+21M

nextstrain_profiles/nextstrain-gisaid-21L/prefilter.smk

@@ -49,7 +49,7 @@ rule gisaid_21L_metadata:
             --exclude \
             --filter-file {input.exclude_clades:q} \
             --key-fields clade \
-            --data-fields Nextstrain_clade \
+            --data-fields clade_nextstrain \
         | sed 1d \
         | zstd -T$(({threads} - 2)) \
         >> {output.metadata:q}