🚧 Read from subsampling config files

nextstrain_profiles/100k/config-gisaid.yaml

@@ -15,7 +15,7 @@ inputs:
 
 builds:
   100k:
-    subsampling_scheme: 100k_scheme
+    subsampling_scheme: subsampling/100k.yaml
 
 # mapping of remote: local files to be uploaded under S3_DST_BUCKET
 upload:
@@ -26,6 +26,7 @@ upload:
 filter:
   exclude_where: "division='USA'"
 
+# FIXME: move this comment to somewhere else then delete the entire entry.
 # We wish to subsample 50k in the previous 12 months and 50k prior to that.
 # Note 1: both --max-date and --min-date are inclusive of the boundary date,
 #         so sequences from that date will be available to both sub-samples

nextstrain_profiles/100k/config-open.yaml

@@ -12,19 +12,9 @@ inputs:
     skip_sanitize_metadata: true
 builds:
   100k:
-    subsampling_scheme: 100k_scheme
+    subsampling_scheme: subsampling/100k.yaml
 upload:
   metadata.tsv.xz: results/100k/100k_subsampled_metadata.tsv.xz
   sequences.fasta.xz: results/100k/100k_subsampled_sequences.fasta.xz
 filter:
   exclude_where: "division='USA'"
-subsampling:
-  100k_scheme:
-    50k_early:
-      group_by: "year month country"
-      max_sequences: 50000
-      max_date: "--max-date 1Y"
-    50k_late:
-      group_by: "year month country"
-      max_sequences: 50000
-      min_date: "--min-date 1Y"

nextstrain_profiles/nextstrain-ci/builds.yaml

@@ -11,27 +11,7 @@ builds:
   # Override the default Nextstrain European build's subsampling scheme for more
   # stable subsampling of a fixed dataset in continuous integration tests.
   europe:
-    subsampling_scheme: nextstrain_ci_sampling
-    region: Europe
-
-subsampling:
-  # Custom subsampling logic for CI tests.
-  nextstrain_ci_sampling:
-    # Focal samples for region
-    region:
-      group_by: "division year month"
-      max_sequences: 20
-      sampling_scheme: "--no-probabilistic-sampling"
-      exclude: "--exclude-where 'region!={region}'"
-    # Contextual samples for region from the rest of the world
-    global:
-      group_by: "year month"
-      max_sequences: 10
-      sampling_scheme: "--no-probabilistic-sampling"
-      exclude: "--exclude-where 'region={region}'"
-      priorities:
-        type: "proximity"
-        focus: "region"
+    subsampling_scheme: subsampling/nextstrain_ci_sampling.yaml
 
 # Override default frequency settings, so we can estimate frequencies from older
 # data with a fixed time range.

nextstrain_profiles/nextstrain-country/builds.yaml

@@ -39,137 +39,22 @@ inputs:
 # Auspice config is specified in rule auspice_config in export_for_nextstrain.smk
 builds:
   nextstrain_country_1m:
-    subsampling_scheme: nextstrain_country_1m
+    subsampling_scheme: subsampling/nextstrain_country_1m.yaml
     title: Genomic epidemiology of SARS-CoV-2 with country-focused subsampling over the past month
-    country: India
   nextstrain_country_2m:
-    subsampling_scheme: nextstrain_country_2m
+    subsampling_scheme: subsampling/nextstrain_country_2m.yaml
     title: Genomic epidemiology of SARS-CoV-2 with country-focused subsampling over the past 2 months
-    country: India
   nextstrain_country_6m:
-    subsampling_scheme: nextstrain_country_6m
+    subsampling_scheme: subsampling/nextstrain_country_6m.yaml
     title: Genomic epidemiology of SARS-CoV-2 with country-focused subsampling over the past 6 months
-    country: India
   nextstrain_country_all-time:
-    subsampling_scheme: nextstrain_country_all-time
+    subsampling_scheme: subsampling/nextstrain_country_all-time.yaml
     title: Genomic epidemiology of SARS-CoV-2 with country-focused subsampling since pandemic start
-    country: India
 
 # remove sequences without division label in US
 filter:
   exclude_where: "division='USA'"
 
-subsampling:
-
-  # Custom subsampling logic for regions over 1m
-  # Grouping by division for North America and Oceania
-  # 4000 total
-  # 4:1 ratio of recent to early
-  # 4:1 ratio of focal to context
-  nextstrain_country_1m:
-    # Early focal samples for region
-    focal_early:
-      group_by: "division year month"
-      max_sequences: 640
-      max_date: "--max-date 1M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_early:
-      group_by: "country year month"
-      max_sequences: 160
-      max_date: "--max-date 1M"
-      exclude: "--exclude-where 'country={country}'"
-    # Recent focal samples for region
-    focal_recent:
-      group_by: "division week"
-      max_sequences: 2560
-      min_date: "--min-date 1M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_recent:
-      group_by: "country week"
-      max_sequences: 640
-      min_date: "--min-date 1M"
-      exclude: "--exclude-where 'country={country}'"
-
-  # Custom subsampling logic for regions over 2m
-  # Grouping by division for North America and Oceania
-  # 4000 total
-  # 4:1 ratio of recent to early
-  # 4:1 ratio of focal to context
-  nextstrain_country_2m:
-    # Early focal samples for region
-    focal_early:
-      group_by: "division year month"
-      max_sequences: 640
-      max_date: "--max-date 2M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_early:
-      group_by: "country year month"
-      max_sequences: 160
-      max_date: "--max-date 2M"
-      exclude: "--exclude-where 'country={country}'"
-    # Recent focal samples for region
-    focal_recent:
-      group_by: "division week"
-      max_sequences: 2560
-      min_date: "--min-date 2M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_recent:
-      group_by: "country week"
-      max_sequences: 640
-      min_date: "--min-date 2M"
-      exclude: "--exclude-where 'country={country}'"
-
-  # Custom subsampling logic for regions over 6m
-  # Grouping by division for North America and Oceania
-  # 4000 total
-  # 4:1 ratio of recent to early
-  # 4:1 ratio of focal to context
-  nextstrain_country_6m:
-    # Early focal samples for region
-    focal_early:
-      group_by: "division year month"
-      max_sequences: 640
-      max_date: "--max-date 6M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_early:
-      group_by: "country year month"
-      max_sequences: 160
-      max_date: "--max-date 6M"
-      exclude: "--exclude-where 'country={country}'"
-    # Recent focal samples for region
-    focal_recent:
-      group_by: "division year month"
-      max_sequences: 2560
-      min_date: "--min-date 6M"
-      exclude: "--exclude-where 'country!={country}'"
-    # Early contextual samples from the rest of the world
-    context_recent:
-      group_by: "country year month"
-      max_sequences: 640
-      min_date: "--min-date 6M"
-      exclude: "--exclude-where 'country={country}'"
-
-  # Custom subsampling logic for regions over all-time
-  # Grouping by division for North America and Oceania
-  # 4000 total
-  # 4:1 ratio of focal to context
-  nextstrain_country_all-time:
-    # Focal samples for country
-    focal:
-      group_by: "division year month"
-      max_sequences: 640
-      exclude: "--exclude-where 'country!={country}'"
-    # Contextual samples from the rest of the world
-    context:
-      group_by: "country year month"
-      max_sequences: 160
-      exclude: "--exclude-where 'country={country}'"
-
 # if different traits should be reconstructed for some builds, specify here
 # otherwise the default trait config in defaults/parameters.yaml will used
 traits: