Merge pull request #1474 from nextstrain/feat/warn-ref-mismatch

packages/nextclade-cli/src/dataset/dataset_download.rs

@@ -12,7 +12,7 @@ use nextclade::io::fasta::{read_one_fasta, read_one_fasta_str};
 use nextclade::io::file::create_file_or_stdout;
 use nextclade::io::fs::{ensure_dir, has_extension, read_file_to_string};
 use nextclade::run::nextclade_wasm::{NextcladeParams, NextcladeParamsOptional};
-use nextclade::tree::tree::AuspiceTree;
+use nextclade::tree::tree::{check_ref_seq_mismatch, AuspiceTree};
 use nextclade::utils::fs::list_files_recursive;
 use nextclade::utils::option::OptionMapRefFallible;
 use nextclade::utils::string::{format_list, surround_with_quotes, Indent};
@@ -143,6 +143,12 @@ pub fn dataset_zip_load(
 
   verify_dataset_files(&virus_properties, zip.file_names());
 
+  if let Some(tree) = &tree {
+    if let Some(tree_ref) = tree.root_sequence() {
+      check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+    }
+  }
+
   Ok(NextcladeParams {
     ref_record,
     gene_map,
@@ -283,6 +289,12 @@ pub fn dataset_dir_load(
     .collect_vec();
   verify_dataset_files(&virus_properties, dataset_dir_files.iter());
 
+  if let Some(tree) = &tree {
+    if let Some(tree_ref) = tree.root_sequence() {
+      check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+    }
+  }
+
   Ok(NextcladeParams {
     ref_record,
     gene_map,
@@ -325,6 +337,12 @@ pub fn dataset_json_load(
       .map_ref_fallible(GeneMap::from_path)
       .wrap_err("When parsing genome annotation")?;
 
+    if let (Some(tree), Some(ref_record)) = (&tree, &ref_record) {
+      if let Some(tree_ref) = tree.root_sequence() {
+        check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+      }
+    }
+
     NextcladeParamsOptional {
       ref_record,
       gene_map,
@@ -370,6 +388,12 @@ pub fn dataset_individual_files_load(
         .map_ref_fallible(AuspiceTree::from_path)
         .wrap_err("When reading reference tree JSON")?;
 
+      if let Some(tree) = &tree {
+        if let Some(tree_ref) = tree.root_sequence() {
+          check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+        }
+      }
+
       Ok(NextcladeParams {
         ref_record,
         gene_map,
@@ -439,6 +463,12 @@ pub fn dataset_str_download_and_load(
     .map_ref_fallible(AuspiceTree::from_str)
     .wrap_err("When reading reference tree from dataset")?;
 
+  if let Some(tree) = &tree {
+    if let Some(tree_ref) = tree.root_sequence() {
+      check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+    }
+  }
+
   Ok(NextcladeParams {
     ref_record,
     gene_map,

packages/nextclade/src/run/nextclade_wasm.rs

@@ -16,7 +16,7 @@ use crate::run::nextclade_run_one::nextclade_run_one;
 use crate::run::params::{NextcladeInputParams, NextcladeInputParamsOptional};
 use crate::translate::translate_genes::Translation;
 use crate::translate::translate_genes_ref::translate_genes_ref;
-use crate::tree::tree::{AuspiceGraph, AuspiceTree, CladeNodeAttrKeyDesc};
+use crate::tree::tree::{check_ref_seq_mismatch, AuspiceGraph, AuspiceTree, CladeNodeAttrKeyDesc};
 use crate::tree::tree_builder::graph_attach_new_nodes_in_place;
 use crate::tree::tree_preprocess::graph_preprocess_in_place;
 use crate::types::outputs::NextcladeOutputs;
@@ -133,6 +133,12 @@ impl NextcladeParams {
             .map_ref_fallible(GeneMap::from_str)
             .wrap_err("When parsing genome annotation")?;
 
+          if let (Some(tree), Some(ref_record)) = (&tree, &ref_record) {
+            if let Some(tree_ref) = tree.root_sequence() {
+              check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+            }
+          }
+
           NextcladeParamsOptional {
             ref_record,
             gene_map,
@@ -160,6 +166,12 @@ impl NextcladeParams {
           .transpose()?
           .unwrap_or_default();
 
+        if let Some(tree) = &tree {
+          if let Some(tree_ref) = tree.root_sequence() {
+            check_ref_seq_mismatch(&ref_record.seq, tree_ref)?;
+          }
+        }
+
         Ok(Self {
           ref_record,
           gene_map,

packages/nextclade/src/tree/tree.rs

@@ -12,6 +12,7 @@ use crate::graph::traits::{HasDivergence, HasName};
 use crate::io::fs::read_file_to_string;
 use crate::io::json::json_parse;
 use eyre::{eyre, Report, WrapErr};
+use log::warn;
 use schemars::JsonSchema;
 use serde::{Deserialize, Deserializer, Serialize, Serializer};
 use std::collections::BTreeMap;
@@ -668,4 +669,37 @@ impl AuspiceTree {
   pub fn map_nodes_mut(&mut self, action: fn((usize, &mut AuspiceTreeNode))) {
     Self::map_nodes_mut_rec(0, &mut self.tree, action);
   }
+
+  pub fn root_sequence(&self) -> Option<&str> {
+    self
+      .root_sequence
+      .as_ref()
+      .and_then(|root_sequence| root_sequence.get("nuc"))
+      .map(String::as_str)
+  }
+}
+
+pub fn check_ref_seq_mismatch(
+  standalone_ref_seq: impl AsRef<str>,
+  tree_ref_seq: impl AsRef<str>,
+) -> Result<(), Report> {
+  if standalone_ref_seq.as_ref() != tree_ref_seq.as_ref() {
+    warn!(
+      r#"Nextclade detected that reference sequence provided does not exactly match reference (root) sequence in Auspice JSON.
+
+     This could be due to one of the reasons:
+
+     - Nextclade dataset author provided reference sequence and reference tree that are incompatible
+     - The reference tree has been constructed incorrectly
+     - The reference sequence provided using `--input-ref` CLI argument is not compatible with the reference tree in the dataset
+     - The reference tree provided using `--input-tree` CLI argument is not compatible with the reference sequence in the dataset
+     - The reference sequence provided using `&input-ref` parameter in Nextclade Web URL is not compatible with the reference tree in the dataset
+     - The reference tree provided using `&input-tree` parameter in Nextclade Web URL is not compatible with the reference sequence in the dataset
+
+     This warning signals that there is a potential for failures if the mismatch is not intended.
+    "#
+    );
+  }
+
+  Ok(())
 }