Add App module and RunSpec-based executor

- Introduce app module with App enum and command adapters
- Add RunSpec to describe image, args and mounts for runs
- Refactor executor to consume RunSpec and pass through to containers
- Extend Docker and HPC paths to use RunSpec
- Implement runners for Score, Rosetta, PyRosetta, RFdiffusion,
  ProteinMPNN, ProteinMPNN Script, LigandMPNN, and Picap

Author: lyskov

src/app.rs

@@ -0,0 +1,95 @@
+use std::collections::HashMap;
+
+use clap::ValueEnum;
+
+mod ligandmpnn;
+mod picap;
+mod proteinmpnn;
+mod proteinmpnn_script;
+mod pyrosetta;
+mod rfdiffusion;
+mod rosetta;
+mod score;
+
+#[derive(ValueEnum, Clone, Copy, Debug, strum::Display)]
+#[clap(rename_all = "lowercase")]
+#[strum(serialize_all = "lowercase")] //  "kebab-case"
+pub enum App {
+    /// Run the Rosetta score command
+    Score,
+
+    /// Run the Rosetta protocol
+    Rosetta,
+
+    /// Start python in env where PyRosetta is installed and execute script
+    #[value(aliases = ["PyRosetta"])]
+    PyRosetta,
+
+    /// Run the RFdiffusion command https://github.com/RosettaCommons/RFdiffusion
+    #[value(aliases = ["Rfdiffusion"])]
+    Rfdiffusion,
+
+    /// Run the ProteinMPNN command https://github.com/dauparas/ProteinMPNN
+    #[value(aliases = ["ProteinMPNN"])]
+    Proteinmpnn,
+
+    /// Run the ProteinMPNN Script command https://github.com/dauparas/ProteinMPNN
+    #[value(aliases = ["Proteinmpnn-script", "ProteinMPNN-Script"])]
+    ProteinmpnnScript,
+
+    /// Run the LigandMPNN command https://github.com/dauparas/LigandMPNN
+    #[value(aliases = ["LigandMPNN"])]
+    Ligandmpnn,
+
+    /// Run the PiCAP/CAPSIF2 command https://github.com/Graylab/picap
+    #[value(aliases = ["PiCAP", "CAPSIF2"])]
+    Picap,
+}
+
+pub struct Image(pub String);
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub enum MountRole {
+    WorkingDir,
+    Scratch,
+}
+
+pub struct RunSpec {
+    pub image: Image,
+    pub args: Vec<String>,
+    //pub scratch: Option<PathBuf>,
+    pub mounts: HashMap<MountRole, String>,
+}
+
+impl RunSpec {
+    pub fn new(image: impl Into<String>, args: Vec<String>) -> Self {
+        Self {
+            image: Image(image.into()),
+            args,
+            mounts: HashMap::new(),
+        }
+    }
+    pub fn scratch(mut self, p: impl Into<String>) -> Self {
+        self.mounts.insert(MountRole::Scratch, p.into());
+        self
+    }
+    pub fn working_dir(mut self, p: impl Into<String>) -> Self {
+        self.mounts.insert(MountRole::WorkingDir, p.into());
+        self
+    }
+}
+
+impl App {
+    pub fn run_spec(self, app_args: Vec<String>) -> RunSpec {
+        match self {
+            App::Score => score::score(app_args),
+            App::Rosetta => rosetta::rosetta(app_args),
+            App::PyRosetta => pyrosetta::pyrosetta(app_args),
+            App::Rfdiffusion => rfdiffusion::rfdiffusion(app_args),
+            App::Proteinmpnn => proteinmpnn::proteinmpnn(app_args),
+            App::ProteinmpnnScript => proteinmpnn_script::proteinmpnn_script(app_args),
+            App::Ligandmpnn => ligandmpnn::ligandmpnn(app_args),
+            App::Picap => picap::picap(app_args),
+        }
+    }
+}

src/app/ligandmpnn.rs

@@ -0,0 +1,13 @@
+use crate::app::RunSpec;
+
+pub fn ligandmpnn(mut app_args: Vec<String>) -> RunSpec {
+    app_args.splice(
+        0..0,
+        [
+            "--out_folder=/w".into(),
+            "--checkpoint_protein_mpnn".into(),
+            "/app/ligandmpnn/model_params/proteinmpnn_v_48_020.pt".into(),
+        ],
+    );
+    RunSpec::new("rosettacommons/ligandmpnn", app_args).working_dir("/w")
+}

src/app/picap.rs

@@ -0,0 +1,13 @@
+use crate::app::RunSpec;
+
+pub fn picap(mut app_args: Vec<String>) -> RunSpec {
+    app_args.splice(
+        0..0,
+        [
+            "--out_folder=/w".into(),
+            "--checkpoint_protein_mpnn".into(),
+            "/app/ligandmpnn/model_params/proteinmpnn_v_48_020.pt".into(),
+        ],
+    );
+    RunSpec::new("rosettacommons/picap", app_args).working_dir("/w")
+}

src/app/proteinmpnn.rs

@@ -0,0 +1,6 @@
+use crate::app::RunSpec;
+
+pub fn proteinmpnn(mut app_args: Vec<String>) -> RunSpec {
+    app_args.splice(0..0, ["--out_folder=/w".into()]);
+    RunSpec::new("rosettacommons/proteinmpnn", app_args).working_dir("/w")
+}

src/app/proteinmpnn_script.rs

@@ -0,0 +1,22 @@
+use crate::app::RunSpec;
+
+pub fn proteinmpnn_script(mut app_args: Vec<String>) -> RunSpec {
+    assert!(
+        app_args.is_empty() || app_args[0].starts_with("-"),
+        "ProteinmpnnScript arguments must include a script name as first argument"
+    );
+
+    let have_input_path_option = ![
+        "helper_scripts/make_bias_AA.py",
+        "helper_scripts/make_pssm_input_dict.py",
+    ]
+    .contains(&app_args[0].as_str());
+
+    if have_input_path_option {
+        app_args.splice(1..1, ["--input_path=/w".into()]);
+    }
+    app_args.splice(1..1, ["--output_path=/w".into()]);
+    app_args[0].insert_str(0, "/app/proteinmpnn/helper_scripts/");
+
+    RunSpec::new("rosettacommons/proteinmpnn", app_args).working_dir("/w")
+}

src/app/pyrosetta.rs

@@ -0,0 +1,6 @@
+use crate::app::RunSpec;
+
+pub fn pyrosetta(mut app_args: Vec<String>) -> RunSpec {
+    app_args.insert(0, "python".into());
+    RunSpec::new("rosettacommons/rosetta:serial", app_args).working_dir("/w")
+}

src/app/rfdiffusion.rs

@@ -0,0 +1,14 @@
+use crate::app::RunSpec;
+
+pub fn rfdiffusion(mut app_args: Vec<String>) -> RunSpec {
+    app_args.splice(
+        0..0,
+        [
+            "inference.output_prefix=/w/".into(),
+            "inference.model_directory_path=/app/RFdiffusion/models".into(),
+        ],
+    );
+    RunSpec::new("rosettacommons/rfdiffusion", app_args)
+        .scratch("/app/RFdiffusion/schedules")
+        .working_dir("/w")
+}

src/app/rosetta.rs

@@ -0,0 +1,5 @@
+use crate::app::RunSpec;
+
+pub fn rosetta(app_args: Vec<String>) -> RunSpec {
+    RunSpec::new("rosettacommons/rosetta:serial", app_args).working_dir("/w")
+}

src/app/score.rs

@@ -0,0 +1,6 @@
+use crate::app::RunSpec;
+
+pub fn score(mut app_args: Vec<String>) -> RunSpec {
+    app_args.insert(0, "score".into());
+    RunSpec::new("rosettacommons/rosetta:serial", app_args).working_dir("/w")
+}