Refinements to phylogenetic build [#2]

* Switch Auspice to default color-by=region * Remove extra `metadata_columns` (and associated config) when calling `augur export v2` * Move `--columns` arg of `augur traits` into config * Add `--metadata-id-columns` arg to `augur traits` invocation -- fixes issue where traits were being improperly inferred * Use a mid-point tree rooting
nextstrain · Jul 31, 2024 · ebf1fde · ebf1fde
1 parent 6f2158b
commit ebf1fde
Show file tree

Hide file tree

Showing 7 changed files with 24 additions and 19 deletions.
diff --git a/phylogenetic/defaults/auspice_config.json b/phylogenetic/defaults/auspice_config.json
@@ -31,6 +31,11 @@
       "key": "country",
       "title": "Country",
       "type": "categorical"
+    },
+    {
+      "key": "host",
+      "title": "Host",
+      "type": "categorical"
     }
   ],
   "geo_resolutions": [
@@ -39,13 +44,14 @@
   ],
   "display_defaults": {
     "map_triplicate": true,
-    "color_by": "clade"
+    "color_by": "region"
   },
   "filters": [
     "clade",
     "region",
     "country",
-    "author"
+    "author",
+    "host"
   ],
   "metadata_columns": [
     "author"

diff --git a/phylogenetic/defaults/config.yaml b/phylogenetic/defaults/config.yaml
@@ -19,5 +19,5 @@ refine:
     clock_filter_iqd: 4
 ancestral:
     inference: "joint"
-export:
-    metadata_columns: "strain division location"
+traits:
+    columns: "region"
diff --git a/phylogenetic/defaults/NC_002031.gb → phylogenetic/defaults/reference.gb b/phylogenetic/defaults/NC_002031.gb → phylogenetic/defaults/reference.gb
diff --git a/phylogenetic/rules/annotate_phylogeny.smk b/phylogenetic/rules/annotate_phylogeny.smk
@@ -57,7 +57,8 @@ rule traits:
     output:
         node_data = "results/{gene}/traits.json",
     params:
-        columns = "region"
+        columns = config["traits"]["columns"],
+        strain_id = config["strain_id_field"],
     log:
         "logs/{gene}/traits.txt",
     benchmark:
@@ -66,9 +67,10 @@ rule traits:
         """
         augur traits \
             --tree {input.tree:q} \
+            --metadata-id-columns {params.strain_id:q} \
             --metadata {input.metadata:q} \
             --output {output.node_data:q} \
-            --columns {params.columns:q} \
+            --columns {params.columns} \
             --confidence \
           2> {log:q}
         """
diff --git a/phylogenetic/rules/construct_phylogeny.smk b/phylogenetic/rules/construct_phylogeny.smk
@@ -48,11 +48,13 @@ rule refine:
         augur refine \
             --tree {input.tree:q} \
             --alignment {input.alignment:q} \
+            --root mid_point \
             --metadata {input.metadata:q} \
             --metadata-id-columns {params.strain_id:q} \
             --output-tree {output.tree:q} \
             --output-node-data {output.node_data:q} \
             --coalescent {params.coalescent:q} \
+            --timetree \
             --date-confidence \
             --date-inference {params.date_inference:q} \
             --clock-filter-iqd {params.clock_filter_iqd:q} \

diff --git a/phylogenetic/rules/export.smk b/phylogenetic/rules/export.smk
@@ -19,7 +19,6 @@ rule export:
         auspice_json = "auspice/yellow-fever-virus_{gene}.json"
     params:
         strain_id = config["strain_id_field"],
-        metadata_columns = config["export"]["metadata_columns"]
     log:
         "logs/{gene}/export.txt",
     benchmark:
@@ -32,7 +31,6 @@ rule export:
             --metadata-id-columns {params.strain_id:q} \
             --node-data {input.branch_lengths:q} {input.traits:q} {input.nt_muts:q} {input.aa_muts:q} \
             --colors {input.colors:q} \
-            --metadata-columns {params.metadata_columns:q} \
             --auspice-config {input.auspice_config:q} \
             --include-root-sequence-inline \
             --output {output.auspice_json:q} \

diff --git a/phylogenetic/rules/prepare_sequences.smk b/phylogenetic/rules/prepare_sequences.smk
@@ -49,24 +49,21 @@ rule filter:
 rule align:
     input:
         sequences="results/genome/filtered.fasta",
-        reference=config["files"]["reference_fasta"],
+        reference=config["files"]["reference_gb"],
         genemap=config["files"]["genemap"],
     output:
         alignment="results/{gene}/aligned.fasta",
-        insertions="results/{gene}/insertions.tsv",
     log:
         "logs/{gene}/align.txt",
     benchmark:
         "benchmarks/{gene}/align.txt"
     shell:
         """
-        (
-          nextclade run \
-              --input-ref {input.reference:q} \
-              --input-annotation {input.genemap:q} \
-              --output-fasta - \
-              --output-tsv {output.insertions:q} \
-              {input.sequences:q} \
-          | seqkit seq -i > {output.alignment:q} \
-        ) 2> {log:q}
+        augur align \
+            --sequences {input.sequences} \
+            --reference-sequence {input.reference} \
+            --output {output.alignment} \
+            --fill-gaps \
+            --remove-reference \
+          2> {log:q}
         """