add command interface to extract list of residues that are in a giv…

…en interface (#4)

* add `list_interface`

* add `test_get_true_interface`

* chore: Update regex dependencies to latest versions

* update README.md

README.md

@@ -6,6 +6,13 @@
 
 A standalone command-line application to generate restraints to be used in HADDOCK.
 
+## Commands
+
+- [`tbl`: Generate TBL file from input file](#tbl-command)
+- [`ti`: Generate true-interface restraints from a PDB file](#ti-command)
+- [`restraint`: Generate Unambiguous restraints to keep molecules together during docking](#restraint-command)
+- [`interface`: List residues in the interface](#interface-command)
+
 ## Planned features
 
 - [x] Generate `.tbl` files from an input file
@@ -16,7 +23,7 @@ A standalone command-line application to generate restraints to be used in HADDO
 - [x] Generate _true-interface_ restraints for benchmarking
 - [x] Create unambiguous restraints to keep molecules together during docking
 - [x] Filter out buried residues
-- [ ] List residues in the interface
+- [x] List residues in the interface
 - [ ] Specify atom subsets
 - [ ] ~Generate random-restraints~ done via CNS
 
@@ -40,10 +47,12 @@ Commands:
   tbl        Generate TBL file from input file
   ti         Generate true-interface restraints from a PDB file
   restraint  Generate Unambiguous restraints to keep molecules together during docking
+  interface  List residues in the interface
   help       Print this message or the help of the given subcommand(s)
 
 Options:
   -h, --help     Print help
+  -V, --version  Print version
 ```
 
 ### `ti` command
@@ -170,6 +179,31 @@ Example:
 ./haddock-restraints restraint examples/2oob_w_gaps.pdb > unambiguous.tbl
 ```
 
+### `interface` command
+
+```bash
+$ ./haddock-restraints interface -h
+List residues in the interface
+
+Usage: haddock-restraints interface <INPUT> <CUTOFF>
+
+Arguments:
+  <INPUT>   PDB file
+  <CUTOFF>  Cutoff distance for interface residues
+
+Options:
+  -h, --help  Print help
+```
+
+Example:
+
+```bash
+./haddock-restraints interface examples/2oob.pdb 5.0
+
+Chain A: [931, 933, 934, 936, 937, 938, 940, 941, 946, 950]
+Chain B: [6, 8, 42, 44, 45, 46, 47, 48, 49, 66, 68, 69, 70]
+```
+
 ***
 
 ## Troubleshooting

src/main.rs

@@ -36,6 +36,13 @@ enum Commands {
         #[arg(help = "PDB file")]
         input: String,
     },
+    #[command(about = "List residues in the interface")]
+    Interface {
+        #[arg(help = "PDB file")]
+        input: String,
+        #[arg(help = "Cutoff distance for interface residues")]
+        cutoff: f64,
+    },
 }
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
@@ -51,6 +58,9 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
         Commands::Restraint { input } => {
             let _ = restraint_bodies(input);
         }
+        Commands::Interface { input, cutoff } => {
+            let _ = list_interface(input, cutoff);
+        }
     }
 
     Ok(())
@@ -180,3 +190,23 @@ fn restraint_bodies(input_file: &str) -> Result<(), Box<dyn Error>> {
 
     Ok(())
 }
+
+fn list_interface(input_file: &str, cutoff: &f64) -> Result<(), Box<dyn Error>> {
+    let pdb = match pdbtbx::open_pdb(input_file, pdbtbx::StrictnessLevel::Loose) {
+        Ok((pdb, _warnings)) => pdb,
+        Err(e) => {
+            panic!("Error opening PDB file: {:?}", e);
+        }
+    };
+
+    let true_interface = structure::get_true_interface(&pdb, *cutoff);
+
+    for (chain_id, residues) in true_interface.iter() {
+        let mut sorted_res = residues.iter().collect::<Vec<_>>();
+        sorted_res.sort();
+
+        println!("Chain {}: {:?}", chain_id, sorted_res);
+    }
+
+    Ok(())
+}

src/structure.rs

@@ -179,3 +179,37 @@ pub fn find_structural_gaps(pdb: &pdbtbx::PDB) -> Vec<Gap> {
 
     gaps
 }
+
+#[cfg(test)]
+mod tests {
+
+    use std::env;
+
+    use super::*;
+
+    #[test]
+    fn test_get_true_interface() {
+        let pdb_path = env::current_dir().unwrap().join("tests/data/complex.pdb");
+
+        let pdb = pdbtbx::open_pdb(pdb_path.to_str().unwrap(), pdbtbx::StrictnessLevel::Loose)
+            .unwrap()
+            .0;
+
+        let observed_true_interface = get_true_interface(&pdb, 5.0);
+
+        let mut expected_true_interface: HashMap<String, HashSet<isize>> = HashMap::new();
+        let chain_a = "A".to_string();
+        let res_a: HashSet<isize> = vec![934, 936, 933, 946, 950, 938, 940, 941, 937, 931]
+            .into_iter()
+            .collect();
+        let chain_b = "B";
+        let res_b: HashSet<isize> = vec![66, 48, 68, 49, 46, 45, 44, 47, 69, 6, 70, 8, 42]
+            .into_iter()
+            .collect();
+
+        expected_true_interface.insert(chain_a, res_a);
+        expected_true_interface.insert(chain_b.to_string(), res_b);
+
+        assert_eq!(observed_true_interface, expected_true_interface);
+    }
+}