feat: add gfa export

Cargo.toml

@@ -48,6 +48,8 @@ eyre = "=0.6.12"
 flate2 = "=1.0.33"
 gcollections = "=1.5.0"
 getset = "0.1.3"
+gfa = { version = "=0.10.1", features = ["serde", "serde_json"] }
+indoc = "=2.0.5"
 intervallum = "=1.4.1"
 itertools = "=0.13.0"
 lazy_static = "=1.5.0"

packages/pangraph/Cargo.toml

@@ -37,6 +37,8 @@ eyre = { workspace = true }
 flate2 = { workspace = true }
 gcollections = { workspace = true }
 getset = { workspace = true }
+gfa = { workspace = true }
+indoc = { workspace = true }
 intervallum = { workspace = true }
 itertools = { workspace = true }
 lazy_static = { workspace = true }

packages/pangraph/src/commands/export/export_args.rs

@@ -56,29 +56,17 @@ pub struct PangraphExportArgs {
   #[clap(value_hint = ValueHint::Other)]
   pub maximum_depth: usize,
 
-  /// Path to directory where output will be stored
-  #[clap(long, short = 'o', default_value = "export")]
-  #[clap(value_hint = ValueHint::DirPath)]
-  pub output_directory: PathBuf,
-
-  /// Basename of files
-  #[clap(long, short = 'p', default_value = "pangraph")]
-  #[clap(value_hint = ValueHint::DirPath)]
-  pub prefix: String,
-
-  /// Do not emit GFA file
-  #[clap(long, alias = "ng")]
-  pub no_export_gfa: bool,
-
-  /// Emit vis directory to input to panX-visualization
-  #[clap(long, alias = "px")]
-  pub export_panx: bool,
-
   /// Do not export any block that contains at least one strain more than once
   #[clap(long, alias = "nd")]
   pub no_duplications: bool,
 
-  /// Random seed
-  #[clap(long)]
-  pub seed: Option<u64>,
+  /// Path to output GFA file
+  #[clap(long, alias = "gfa")]
+  #[clap(value_hint = ValueHint::Other)]
+  pub output_gfa: Option<PathBuf>,
+
+  /// Path to output directory where PanX visualization files will be written
+  #[clap(long, alias = "panx")]
+  #[clap(value_hint = ValueHint::DirPath)]
+  pub output_panx: Option<PathBuf>,
 }

packages/pangraph/src/commands/export/export_run.rs

@@ -1,9 +1,10 @@
 use crate::commands::export::export_args::PangraphExportArgs;
+use crate::io::gfa::gfa_write_file;
+use crate::make_error;
 use crate::pangraph::pangraph::Pangraph;
-use crate::utils::random::get_random_number_generator;
 use eyre::Report;
 
-pub fn export_run(args: &PangraphExportArgs) -> Result<(), Report> {
+pub fn export_run(args: PangraphExportArgs) -> Result<(), Report> {
   let PangraphExportArgs {
     input_json,
     edge_minimum_length,
@@ -14,17 +15,25 @@ pub fn export_run(args: &PangraphExportArgs) -> Result<(), Report> {
     maximum_length,
     minimum_depth,
     maximum_depth,
-    output_directory,
-    prefix,
-    no_export_gfa,
-    export_panx,
     no_duplications,
-    seed,
-  } = &args;
+    output_gfa,
+    output_panx,
+  } = args;
 
-  let rng = get_random_number_generator(seed);
+  if [&output_gfa, &output_panx].iter().all(|o| o.is_none()) {
+    return make_error!("No output formats specified. Specify at least one output path.");
+  }
 
-  let pangraph_json = Pangraph::from_path(input_json)?;
+  let graph = Pangraph::from_path(&input_json)?;
+
+  if let Some(output_gfa) = output_gfa {
+    gfa_write_file(output_gfa, &graph)?;
+  }
+
+  if let Some(output_panx) = output_panx {
+    // TODO
+    // panx_write(output_panx, &graph)?;
+  }
 
   Ok(())
 }

packages/pangraph/src/commands/main.rs

@@ -17,7 +17,7 @@ pub fn pangraph_main() -> Result<(), Report> {
 
   match args.command {
     PangraphCommands::Build(args) => build_run(&args),
-    PangraphCommands::Export(args) => export_run(&args),
+    PangraphCommands::Export(args) => export_run(args),
     PangraphCommands::Simplify(args) => simplify_run(args),
     PangraphCommands::Reconstruct(args) => reconstruct_run(&args),
     PangraphCommands::Schema(args) => generate_schema(&args),