Merge pull request #910 from nextstrain/multiple-timespans

Split GISAID profile to "six-month" and "all-time" builds

.github/workflows/rebuild-gisaid.yml

@@ -50,14 +50,14 @@ jobs:
         nextstrain build \
           --aws-batch \
           --detach \
-          --cpus 36 \
-          --memory 70GiB \
+          --cpus 72 \
+          --memory 140GiB \
           . \
             deploy \
             upload \
             --config "${config[@]}" \
             --profile nextstrain_profiles/nextstrain-gisaid \
-            --set-threads tree=16 \
+            --set-threads tree=8 \
         |& tee build-launch.log
       env:
         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}

.github/workflows/rebuild-open.yml

@@ -51,14 +51,14 @@ jobs:
         nextstrain build \
           --aws-batch \
           --detach \
-          --cpus 36 \
-          --memory 70GiB \
+          --cpus 72 \
+          --memory 140GiB \
           . \
             deploy \
             upload \
             --config "${config[@]}" \
             --profile nextstrain_profiles/nextstrain-open \
-            --set-threads tree=16 \
+            --set-threads tree=8 \
         |& tee build-launch.log
       env:
         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}

defaults/parameters.yaml

@@ -81,8 +81,8 @@ filter:
   exclude_where: "division='USA'"
   exclude_ambiguous_dates_by: "any"
 
-  # Exclude sequences which are from before late 2019 (likely date mix-ups)
-  min_date: 2019.74
+  # Exclude sequences which are from before Dec 2019 (likely date mix-ups)
+  min_date: "2019-12-01"
 
 # When choosing contextual samples for a focal set, applying crowding penalty
 # will help reduce the number of genetically identical strains that get chosen,
@@ -136,10 +136,10 @@ frequencies:
   # min_date is set by default to 1 year before present
   # but can be explicitly set if desired
 
-  # Number of months between pivots
+  # Number of weeks between pivots
   pivot_interval: 1
 
-  # Measure pivots in weeks rather than months
+  # Measure pivots in weeks
   pivot_interval_units: "weeks"
 
   # KDE bandwidths in proportion of a year to use per strain.

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
 
+- 29 April 2022: Include multiple timespans in Nextstrain profile builds. [PR 910](https://github.com/nextstrain/ncov/pull/910)
+
 - 29 April 2022: Update default mask parameters to mask 200 bases from the end of the genome rather than the existing 50. This was necessary because there is a large deletion in this region in circulating 21L viruses. This deletion is causing problems with alignment and the resulting mis-alignment appears as excess mutations in the tree. [PR 939](https://github.com/nextstrain/ncov/pull/939).
 
 - 27 April 2022: Include new clades 22A, 22B and 22C, where 22A corresponds to Pango lineage BA.4, 22B corresponds to Pango lineage BA.5 and 22C corresponds to Pango lineage BA.2.12.1. Please see [PR 933](https://github.com/nextstrain/ncov/pull/933) for rationale behind these clade updates.