feat: output Genebank feature table file

packages_rs/nextclade-cli/src/cli/nextalign_cli.rs

@@ -67,6 +67,7 @@ pub enum NextalignOutputSelection {
   All,
   Fasta,
   Translations,
+  FeatureTable,
   Insertions,
   Errors,
 }
@@ -148,7 +149,7 @@ pub struct NextalignRunOutputArgs {
   ///
   /// If both the `--output-all` and individual `--output-*` flags are provided, each individual flag overrides the corresponding default output path.
   ///
-  /// At least one of the output flags is required: `--output-all`, `--output-fasta`, `--output-translations`, `--output-insertions`, `--output-errors`
+  /// At least one of the output flags is required: `--output-all`, `--output-fasta`, `--output-translations`, `--output-feature-table`, `--output-insertions`, `--output-errors`
   ///
   /// If the required directory tree does not exist, it will be created.
   #[clap(long, short = 'O')]
@@ -212,6 +213,22 @@ pub struct NextalignRunOutputArgs {
   #[clap(value_hint = ValueHint::AnyPath)]
   pub output_translations: Option<String>,
 
+  /// Path to output Genbank Feature Table file (.tbl).
+  ///
+  /// Writes Feature Table in Genbank file format, which can be used to facilitate submissions to Genbank database.
+  ///
+  /// See: https://www.ncbi.nlm.nih.gov/genbank/feature_table/
+  ///
+  /// Takes precedence over paths configured with `--output-all`, `--output-basename` and `--output-selection`.
+  ///
+  /// If filename ends with one of the supported file extensions: `gz`, `bz2`, `xz`, `zstd`, it will be transparently
+  /// compressed. If a filename is "-" then the output will be written uncompressed to standard output (stdout).
+  ///
+  /// If the required directory tree does not exist, it will be created.
+  #[clap(long, short = 'F')]
+  #[clap(value_hint = ValueHint::AnyPath)]
+  pub output_feature_table: Option<PathBuf>,
+
   /// Path to output CSV file that contain insertions stripped from the reference alignment.
   ///
   /// Takes precedence over paths configured with `--output-all`, `--output-basename` and `--output-selection`.
@@ -320,6 +337,7 @@ pub fn nextalign_get_output_filenames(run_args: &mut NextalignRunArgs) -> Result
         output_selection,
         output_fasta,
         output_translations,
+        output_feature_table,
         output_insertions,
         output_errors,
         include_reference,
@@ -373,6 +391,10 @@ pub fn nextalign_get_output_filenames(run_args: &mut NextalignRunArgs) -> Result
         output_translations.get_or_insert(output_translations_template)
       };
     }
+
+    if output_selection.contains(&NextalignOutputSelection::FeatureTable) {
+      let output_feature_table = output_feature_table.get_or_insert(add_extension(&default_output_file_path, "tbl"));
+    }
   }
 
   if let Some(output_translations) = output_translations {

packages_rs/nextclade-cli/src/cli/nextalign_loop.rs

@@ -40,6 +40,7 @@ pub fn nextalign_run(run_args: NextalignRunArgs) -> Result<(), Report> {
         output_selection,
         output_fasta,
         output_translations,
+        output_feature_table,
         output_insertions,
         output_errors,
         include_reference,
@@ -152,6 +153,7 @@ pub fn nextalign_run(run_args: NextalignRunArgs) -> Result<(), Report> {
         &gene_map,
         &output_fasta,
         &output_translations,
+        &output_feature_table,
         &output_insertions,
         &output_errors,
         in_order,

packages_rs/nextclade-cli/src/cli/nextalign_ordered_writer.rs

@@ -3,6 +3,7 @@ use eyre::{Report, WrapErr};
 use log::warn;
 use nextclade::io::errors_csv::ErrorsCsvWriter;
 use nextclade::io::fasta::{FastaPeptideWriter, FastaRecord, FastaWriter};
+use nextclade::io::genbank_feature_table::{GenbankFeatureTableEntry, GenbankFeatureTableFileWriter};
 use nextclade::io::gene_map::GeneMap;
 use nextclade::io::insertions_csv::InsertionsCsvWriter;
 use nextclade::io::nuc::from_nuc_seq;
@@ -17,6 +18,7 @@ use std::path::PathBuf;
 pub struct NextalignOrderedWriter<'a> {
   fasta_writer: Option<FastaWriter>,
   fasta_peptide_writer: Option<FastaPeptideWriter>,
+  feature_table_writer: Option<GenbankFeatureTableFileWriter<'a>>,
   insertions_csv_writer: Option<InsertionsCsvWriter>,
   errors_csv_writer: Option<ErrorsCsvWriter<'a>>,
   expected_index: usize,
@@ -29,6 +31,7 @@ impl<'a> NextalignOrderedWriter<'a> {
     gene_map: &'a GeneMap,
     output_fasta: &Option<PathBuf>,
     output_translations: &Option<String>,
+    output_feature_table: &Option<PathBuf>,
     output_insertions: &Option<PathBuf>,
     output_errors: &Option<PathBuf>,
     in_order: bool,
@@ -38,6 +41,9 @@ impl<'a> NextalignOrderedWriter<'a> {
     let fasta_peptide_writer = output_translations
       .map_ref_fallible(|output_translations| FastaPeptideWriter::new(gene_map, &output_translations))?;
 
+    let feature_table_writer = output_feature_table
+      .map_ref_fallible(|output_feature_table| GenbankFeatureTableFileWriter::new(output_feature_table, gene_map))?;
+
     let insertions_csv_writer = output_insertions.map_ref_fallible(InsertionsCsvWriter::new)?;
 
     let errors_csv_writer =
@@ -46,6 +52,7 @@ impl<'a> NextalignOrderedWriter<'a> {
     Ok(Self {
       fasta_writer,
       fasta_peptide_writer,
+      feature_table_writer,
       insertions_csv_writer,
       errors_csv_writer,
       expected_index: 0,
@@ -85,6 +92,7 @@ impl<'a> NextalignOrderedWriter<'a> {
           stripped,
           alignment,
           translations,
+          gene_ranges_qry,
           warnings,
           missing_genes,
           is_reverse_complement,
@@ -100,6 +108,13 @@ impl<'a> NextalignOrderedWriter<'a> {
           }
         }
 
+        if let Some(feature_table_writer) = &mut self.feature_table_writer {
+          feature_table_writer.write(&GenbankFeatureTableEntry {
+            seq_name: seq_name.clone(),
+            gene_ranges_qry: gene_ranges_qry.clone(),
+          })?;
+        }
+
         if let Some(insertions_csv_writer) = &mut self.insertions_csv_writer {
           insertions_csv_writer.write(&seq_name, &stripped.insertions, translations)?;
         }

packages_rs/nextclade-cli/src/cli/nextclade_cli.rs

@@ -214,6 +214,7 @@ pub enum NextcladeOutputSelection {
   Tsv,
   Tree,
   Translations,
+  FeatureTable,
   Insertions,
   Errors,
 }
@@ -361,7 +362,7 @@ pub struct NextcladeRunOutputArgs {
   ///
   /// If both the `--output-all` and individual `--output-*` flags are provided, each individual flag overrides the corresponding default output path.
   ///
-  /// At least one of the output flags is required: `--output-all`, `--output-fasta`, `--output-ndjson`, `--output-json`, `--output-csv`, `--output-tsv`, `--output-tree`, `--output-translations`, `--output-insertions`, `--output-errors`
+  /// At least one of the output flags is required: `--output-all`, `--output-fasta`, `--output-ndjson`, `--output-json`, `--output-csv`, `--output-tsv`, `--output-tree`, `--output-translations`, `--output-feature-table`, `--output-insertions`, `--output-errors`
   ///
   /// If the required directory tree does not exist, it will be created.
   #[clap(long, short = 'O')]
@@ -425,6 +426,22 @@ pub struct NextcladeRunOutputArgs {
   #[clap(value_hint = ValueHint::AnyPath)]
   pub output_translations: Option<String>,
 
+  /// Path to output Genbank Feature Table file (.tbl).
+  ///
+  /// Writes Feature Table in Genbank file format, which can be used to facilitate submissions to Genbank database.
+  ///
+  /// See: https://www.ncbi.nlm.nih.gov/genbank/feature_table/
+  ///
+  /// Takes precedence over paths configured with `--output-all`, `--output-basename` and `--output-selection`.
+  ///
+  /// If filename ends with one of the supported file extensions: `gz`, `bz2`, `xz`, `zstd`, it will be transparently
+  /// compressed. If a filename is "-" then the output will be written uncompressed to standard output (stdout).
+  ///
+  /// If the required directory tree does not exist, it will be created.
+  #[clap(long, short = 'F')]
+  #[clap(value_hint = ValueHint::AnyPath)]
+  pub output_feature_table: Option<PathBuf>,
+
   /// Path to output Newline-delimited JSON (NDJSON) results file.
   ///
   /// This file format is most suitable for further machine processing of the results. By contrast to plain json, it can be streamed line-by line, so much bigger outputs are feasible.
@@ -615,6 +632,7 @@ pub fn nextclade_get_output_filenames(run_args: &mut NextcladeRunArgs) -> Result
         output_selection,
         output_fasta,
         output_translations,
+        output_feature_table,
         output_ndjson,
         output_json,
         output_csv,
@@ -671,6 +689,10 @@ pub fn nextclade_get_output_filenames(run_args: &mut NextcladeRunArgs) -> Result
       output_translations.get_or_insert(output_translations_template);
     }
 
+    if output_selection.contains(&NextcladeOutputSelection::FeatureTable) {
+      let output_feature_table = output_feature_table.get_or_insert(add_extension(&default_output_file_path, "tbl"));
+    }
+
     if output_selection.contains(&NextcladeOutputSelection::Ndjson) {
       output_ndjson.get_or_insert(add_extension(&default_output_file_path, "ndjson"));
     }

packages_rs/nextclade-cli/src/cli/nextclade_loop.rs

@@ -7,6 +7,7 @@ use crate::dataset::dataset_download::{
 };
 use crossbeam::thread;
 use eyre::{Report, WrapErr};
+use indexmap::IndexMap;
 use itertools::Itertools;
 use log::info;
 use nextclade::align::gap_open::{get_gap_open_close_scores_codon_aware, get_gap_open_close_scores_flat};
@@ -22,12 +23,13 @@ use nextclade::translate::translate_genes_ref::translate_genes_ref;
 use nextclade::tree::tree_attach_new_nodes::tree_attach_new_nodes_in_place;
 use nextclade::tree::tree_preprocess::tree_preprocess_in_place;
 use nextclade::types::outputs::NextcladeOutputs;
+use nextclade::utils::range::Range;
 use std::path::PathBuf;
 
 pub struct NextcladeRecord {
   pub index: usize,
   pub seq_name: String,
-  pub outputs_or_err: Result<(Vec<Nuc>, Vec<Translation>, NextcladeOutputs), Report>,
+  pub outputs_or_err: Result<(Vec<Nuc>, Vec<Translation>, IndexMap<String, Range>, NextcladeOutputs), Report>,
 }
 
 pub struct DatasetFilePaths {
@@ -83,6 +85,7 @@ pub fn nextclade_run(run_args: NextcladeRunArgs) -> Result<(), Report> {
         output_selection,
         output_fasta,
         output_translations,
+        output_feature_table,
         output_ndjson,
         output_json,
         output_csv,
@@ -229,6 +232,7 @@ pub fn nextclade_run(run_args: NextcladeRunArgs) -> Result<(), Report> {
         &output_insertions,
         &output_errors,
         &output_translations,
+        &output_feature_table,
         in_order,
       )
       .wrap_err("When creating output writer")
@@ -243,7 +247,7 @@ pub fn nextclade_run(run_args: NextcladeRunArgs) -> Result<(), Report> {
 
       for record in result_receiver {
         if should_keep_outputs {
-          if let Ok((_, _, nextclade_outputs)) = &record.outputs_or_err {
+          if let Ok((_, _, _, nextclade_outputs)) = &record.outputs_or_err {
             outputs.push(nextclade_outputs.clone());
           }
         }

packages_rs/nextclade-cli/src/cli/nextclade_ordered_writer.rs

@@ -4,6 +4,7 @@ use itertools::Itertools;
 use log::warn;
 use nextclade::io::errors_csv::ErrorsCsvWriter;
 use nextclade::io::fasta::{FastaPeptideWriter, FastaRecord, FastaWriter};
+use nextclade::io::genbank_feature_table::{GenbankFeatureTableEntry, GenbankFeatureTableFileWriter};
 use nextclade::io::gene_map::GeneMap;
 use nextclade::io::insertions_csv::InsertionsCsvWriter;
 use nextclade::io::ndjson::NdjsonFileWriter;
@@ -23,6 +24,7 @@ use std::path::PathBuf;
 pub struct NextcladeOrderedWriter<'a> {
   fasta_writer: Option<FastaWriter>,
   fasta_peptide_writer: Option<FastaPeptideWriter>,
+  feature_table_writer: Option<GenbankFeatureTableFileWriter<'a>>,
   output_json_writer: Option<ResultsJsonWriter>,
   output_ndjson_writer: Option<NdjsonFileWriter>,
   output_csv_writer: Option<NextcladeResultsCsvFileWriter>,
@@ -46,13 +48,17 @@ impl<'a> NextcladeOrderedWriter<'a> {
     output_insertions: &Option<PathBuf>,
     output_errors: &Option<PathBuf>,
     output_translations: &Option<String>,
+    output_feature_table: &Option<PathBuf>,
     in_order: bool,
   ) -> Result<Self, Report> {
     let fasta_writer = output_fasta.map_ref_fallible(FastaWriter::from_path)?;
 
     let fasta_peptide_writer = output_translations
       .map_ref_fallible(|output_translations| FastaPeptideWriter::new(gene_map, &output_translations))?;
 
+    let feature_table_writer = output_feature_table
+      .map_ref_fallible(|output_feature_table| GenbankFeatureTableFileWriter::new(output_feature_table, gene_map))?;
+
     let insertions_csv_writer = output_insertions.map_ref_fallible(InsertionsCsvWriter::new)?;
 
     let errors_csv_writer =
@@ -75,6 +81,7 @@ impl<'a> NextcladeOrderedWriter<'a> {
     Ok(Self {
       fasta_writer,
       fasta_peptide_writer,
+      feature_table_writer,
       output_json_writer,
       output_ndjson_writer,
       output_csv_writer,
@@ -113,7 +120,7 @@ impl<'a> NextcladeOrderedWriter<'a> {
     } = record;
 
     match outputs_or_err {
-      Ok((qry_seq_stripped, translations, nextclade_outputs)) => {
+      Ok((qry_seq_stripped, translations, gene_ranges_qry, nextclade_outputs)) => {
         let NextcladeOutputs {
           warnings,
           insertions,
@@ -132,6 +139,13 @@ impl<'a> NextcladeOrderedWriter<'a> {
           }
         }
 
+        if let Some(feature_table_writer) = &mut self.feature_table_writer {
+          feature_table_writer.write(&GenbankFeatureTableEntry {
+            seq_name: seq_name.clone(),
+            gene_ranges_qry: gene_ranges_qry.clone(),
+          })?;
+        }
+
         if let Some(insertions_csv_writer) = &mut self.insertions_csv_writer {
           insertions_csv_writer.write(&seq_name, insertions, &translations)?;
         }

packages_rs/nextclade-web/src/wasm/analyze.rs

@@ -78,7 +78,7 @@ impl NextcladeParams {
 pub struct AnalysisInput {
   #[wasm_bindgen(getter_with_clone)]
   pub qry_index: usize,
-  
+
   #[wasm_bindgen(getter_with_clone)]
   pub qry_seq_name: String,
 
@@ -243,7 +243,7 @@ impl Nextclade {
       &self.gap_open_close_aa,
       &self.aln_params,
     ) {
-      Ok((qry_seq_aligned_stripped, translations, nextclade_outputs)) => {
+      Ok((qry_seq_aligned_stripped, translations, gene_ranges_qry, nextclade_outputs)) => {
         let nextclade_outputs_str =
           json_stringify(&nextclade_outputs).wrap_err("When serializing output results of Nextclade")?;
 

packages_rs/nextclade/Cargo.toml

@@ -29,10 +29,11 @@ env_logger = "0.9.0"
 eyre = "0.6.8"
 flate2 = "1.0.24"
 getrandom = "0.2.6"
-indexmap = { version = "1.8.1", features = ["serde"] }
+indexmap = { version = "1.8.1", features = ["serde-1", "rayon"] }
 itertools = "0.10.3"
 lazy_static = "1.4.0"
 log = "0.4.16"
+multimap = "0.8.3"
 num = "0.4.0"
 num-traits = "0.2.14"
 num_cpus = "1.13.1"

packages_rs/nextclade/src/gene/gene.rs

@@ -1,5 +1,21 @@
 use crate::utils::range::Range;
+use multimap::MultiMap;
 use serde::{Deserialize, Serialize};
+use std::str::FromStr;
+use strum_macros::{EnumString, Display as EnumDisplay};
+
+#[derive(Clone, Debug, Deserialize, Serialize, Eq, PartialEq, EnumString, EnumDisplay)]
+pub enum FeatureKind {
+  #[strum(serialize = "gene")]
+  #[serde(rename = "gene")]
+  Gene,
+
+  CDS,
+
+  #[strum(serialize = "unknown")]
+  #[serde(rename = "unknown")]
+  Unknown,
+}
 
 #[derive(Clone, Debug, Deserialize, Serialize, Eq, PartialEq)]
 pub enum GeneStrand {
@@ -24,11 +40,13 @@ impl From<bio_types::strand::Strand> for GeneStrand {
 #[derive(Clone, Debug, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
 pub struct Gene {
+  pub kind: FeatureKind,
   pub gene_name: String,
   pub start: usize,
   pub end: usize,
   pub strand: GeneStrand,
   pub frame: i32,
+  pub attributes: MultiMap<String, String>,
 }
 
 impl Gene {