Merge pull request #1555 from nextstrain/fix-titer-sub-tree-annotations

Fix titer substitution model tree annotations

CHANGES.md

@@ -8,8 +8,13 @@
 * traits, export v2: `augur traits` now reports all confidence values above 0.1% rather than limiting them to the top 4 results. There is no change in the eventual Auspice dataset as `augur export v2` will still only consider the top 4. [#1512][] (@jameshadfield)
 * curate: Excel (`.xlsx` and `.xls`) and OpenOffice (`.ods`) spreadsheet files are now also supported as metadata inputs (`--metadata`).  The first sheet in the workbook is read as tabular data.  [#1550][] (@tsibley)
 
+### Bug Fixes
+
+* titers sub: Fixes a bug where antigenic weights were assigned to branches for substitutions in the incorrect order of `<derived allele><position><ancestral allele>` instead of `<ancestral allele><position><derived allele>`. [#1555][] (@huddlej)
+
 [#1512]: https://github.com/nextstrain/augur/pull/1512
 [#1550]: https://github.com/nextstrain/augur/pull/1550
+[#1555]: https://github.com/nextstrain/augur/pull/1555
 
 
 ## 25.1.1 (15 July 2024)

augur/titer_model.py

@@ -1142,7 +1142,7 @@ def annotate_tree(self, tree):
         for node in tree.find_clades():
             for child in node.clades:
                 # Get mutations between the current node and its parent.
-                mutations = self.get_mutations(child.name, node.name)
+                mutations = self.get_mutations(node.name, child.name)
 
                 # Calculate titer drop on the branch to the current node.
                 child.dTiterSub = 0

tests/functional/titers/cram/titers-sub-with-tree.t

@@ -17,3 +17,11 @@ Test titer substitution model with alignment and tree inputs.
    --- 272 total measurements
   $ grep cTiterSub $TMP/titers-sub.json | wc -l
   \s*120 (re)
+
+Verify that the titer drops assigned per branch correspond to the expected values for this dataset.
+In this example, we know that the HA1 amino acid sequence for A/Fujian/445/2003 carries a S193N substitution and that the titer model assigns a weight of 0.6 to that substitution.
+The titer model assigns a higher weight of 1.22 to the opposite substitution N193S.
+When we search for that sequence's per-branch titer drop, we should get the smaller value below.
+
+  $ python3 -c 'import json, sys; print(json.load(sys.stdin)["nodes"]["A/Fujian/445/2003"]["dTiterSub"])' < $TMP/titers-sub.json
+  0.60* (glob)