Moved Newick C code to new file. Added code that will eventually crea…

…te a Tree struct from a rapidNJ newick string. I do this by modifying the rapidNJ Newick string and passing it to the C code. Currently the C code throws an error

build.rs

@@ -1,10 +1,10 @@
 fn main() {
-    cxx_build::bridge("src/phylo2vec.rs")
+    cxx_build::bridge("src/newick.rs")
         .file("src/phylo2vec.cpp")
         .std("c++17")
         .compile("phylo2vec");
 
-        println!("cargo:rerun-if-changed=src/phylo2vec.rs");
+        println!("cargo:rerun-if-changed=src/newick.rs");
         println!("cargo:rerun-if-changed=src/phylo2vec.cpp");
         println!("cargo:rerun-if-changed=src/phylo2vec.hpp");
 }

src/lib.rs

@@ -6,6 +6,7 @@ mod phylo2vec;
 mod tests;
 mod tree;
 mod dspsa;
+mod newick;
 
 use cxx::kind;
 use cxx::let_cxx_string;
@@ -16,6 +17,8 @@ use regex::{RegexSet, RegexSetBuilder};
 
 use crate::dspsa::hill_peturb;
 use crate::gen_list::*;
+use crate::newick::newick_to_vec;
+use crate::newick::parse_rapidNJ_newick;
 use crate::phylo2vec::*;
 use crate::tree::Tree;
 use crate::likelihoods::logse;
@@ -33,51 +36,9 @@ pub fn main() {
     let args = cli_args();
     // Newick string example
     let nstr: String = String::from("(((((('9':0.00066056,'4':0.0012556):0.00030238,('25':0.00042479,'19':0.00030083):0.00039731):0.00012351,'7':0.00039526):8.5797e-05,'18':0.00068249):0.00011977,'20':0.00056004):9.6199e-05,(((((((('17':0.0045802,'8':0.0021951):0.00019116,(((('24':0.0021162,'12':0.0017684):0.00074803,'13':0.0043482):0.00026831,('27':0.00063335,'5':0.00071804):0.0024273):4.8521e-05,((('23':0.0012435,'2':0.0012676):0.00011672,('21':0.00085695,'6':0.00096101):0.001509):0.00046989,'10':0.0020611):0.00092726):0.00022606):0.00019029,'26':0.0027488):0.00030355,('14':0.0021462,'0':0.0018751):0.00061029):0.00035469,'1':0.00096653):0.00015559,'22':0.0013558):7.0943e-05,('16':0.00054383,'15':0.00076666):0.00015841):0.00016779,'11':0.0021712):9.9157e-05,'3':0.00081132);");
-
-    // Chops into nodes
-    let mut v: Vec<&str> = nstr.split_inclusive(|c: char| c.eq(&')') || c.eq(&',')).filter(|s| !s.is_empty() && s.ne(&";")).collect();
-    // Pop off the semi-colon
-    // v.pop();
-    // println!("{:?}", v);
-
-    let n_nodes = v.len();
-    let mut branch_len: Vec<f64> = Vec::with_capacity(n_nodes);
-    let mut name_dict: HashMap<usize, String> = HashMap::with_capacity(n_nodes);
-    let internal_index = ((n_nodes - 1) / 2) + 1;
-    let mut node_index: usize = 0;
-
-    // println!("{}", n_nodes);
-    // println!("{}", internal_index);
-
-    // Iterate over nodes
-    for node in v {
-        let cleaned_node: String = node.replace(&['(', '\'', ')', ','], "");
-        // println!("{:?}", cleaned_node);
-        let split_node: Vec<&str> = cleaned_node.split(':').collect();
-        // println!("{:?}", split_node);
-
-        let node_name: String = match split_node[0].is_empty() {
-            true => node_index.to_string(),
-            false => split_node[0].to_string(),
-        };
-        name_dict.insert(node_index, node_name);
-
-        let blen: f64 = split_node[1].parse().unwrap();
-        branch_len.push(blen);
-        node_index += 1;
+    let x: String = parse_rapidNJ_newick(&nstr);
+    // println!("{:?}", newick_to_vec(&x, 27));    
 
-        // Need to reconstruct new newick string using new node labels
-
-        // let new_el: Vec<&str> = node.split_inclusive(|c: char| c.eq(&')') || c.eq(&',') || c.eq(&'(') || c.eq(&':') || c.eq(&',')).collect();
-
-        // println!("{:?}", node);
-    }
-
-    // for (key, val) in name_dict.iter() {
-    //     println!("key: {key} val: {val}");
-    // }
-    // let v = newick_to_vec(&nstr, 27);
-
 
     // Define rate matrix
     let q: na::Matrix4<f64> = na::Matrix4::new(

src/newick.rs

@@ -0,0 +1,74 @@
+use cxx::let_cxx_string;
+use std::collections::HashMap;
+
+#[cxx::bridge]
+pub mod ffi {
+    unsafe extern "C++" {
+        include!("bactrees/include/phylo2vec.hpp");
+        fn doToVector(newick: Pin<&mut CxxString>, num_leaves: i32, with_mapping: bool) -> UniquePtr<CxxVector<i32>>;
+    }
+}
+
+pub fn newick_to_vec(nw: &String, n_leaves: usize) -> Vec<usize>{
+    let_cxx_string!(nw_cpp = nw);
+    let x = ffi::doToVector(nw_cpp, n_leaves as i32, false);
+    let y: Vec<usize> = x.iter().map(|el| *el as usize).collect();
+    y
+}
+
+pub fn parse_rapidNJ_newick(nw: &String) -> String {
+
+    // Chop the Newick string into nodes
+    let v: Vec<&str> = nw.split_inclusive(|c: char| c.eq(&')') || c.eq(&','))
+    .filter(|s| !s.is_empty() && s.ne(&";"))
+    .collect();
+
+    // Initiate data structures
+    let n_nodes = v.len();
+    let mut branch_len: Vec<f64> = Vec::with_capacity(n_nodes);
+    let mut name_dict: HashMap<usize, String> = HashMap::with_capacity(n_nodes);
+    let mut node_index: usize = 0;
+    let mut new_newick: Vec<String> = Vec::with_capacity(n_nodes);
+
+    // Iterate over nodes
+    for node in v {
+        // Cut outbrackets and commas
+        let cleaned_node: String = node.replace(&['(', '\'', ')', ','], "");
+
+        // Split into label and branch length
+        let split_node: Vec<&str> = cleaned_node.split(':').collect();
+
+        // Get old node label (an old label is assigned if no previous 
+        // label due to being internal node)
+        let node_name: String = match split_node[0].is_empty() {
+            true => node_index.to_string(),
+            false => split_node[0].to_string(),
+        };
+        // Put into name HashMap
+        name_dict.insert(node_index, node_name);
+
+        // Get branch length
+        let blen: f64 = split_node[1].parse().unwrap();
+        branch_len.push(blen);
+
+        // Put new node labels into old newick string
+        if split_node[0].is_empty() {
+            let new_el: String = node.replace(":", &format!("{}{}", node_index.to_string(), ":"));
+            new_newick.push(new_el);
+
+        } else {
+            let new_el: String = node.replace(&format!("'{}'", name_dict.get(&node_index).unwrap()), 
+            &format!("{}", node_index.to_string()));
+            new_newick.push(new_el);
+        }
+
+        // Iterate node index
+        node_index += 1;
+    }
+
+    // Put newick string with new labels together
+    let mut new_str: String = new_newick.join("");
+    new_str.push_str(";");
+
+    new_str
+}

src/phylo2vec.rs

@@ -1,7 +1,6 @@
 use crate::Tree;
 use ndarray::*;
 use rand::{seq::SliceRandom, thread_rng, Rng};
-use cxx::let_cxx_string;
 
 pub fn phylo2vec_quad(v: &[usize]) -> Tree {
     let mut tree = Tree::new(v);
@@ -214,19 +213,4 @@ impl Tree {
 
     }
 
-}
-
-#[cxx::bridge]
-pub mod ffi {
-    unsafe extern "C++" {
-        include!("bactrees/include/phylo2vec.hpp");
-        fn doToVector(newick: Pin<&mut CxxString>, num_leaves: i32, with_mapping: bool) -> UniquePtr<CxxVector<i32>>;
-    }
-}
-
-pub fn newick_to_vec(nw: &String, n_leaves: usize) -> Vec<usize>{
-    let_cxx_string!(nw_cpp = nw);
-    let x = ffi::doToVector(nw_cpp, n_leaves as i32, false);
-    let y: Vec<usize> = x.iter().map(|el| *el as usize).collect();
-    y
 }

src/tests.rs

@@ -4,7 +4,7 @@ mod tests {
     use crate::phylo2vec::phylo2vec_lin;
     use crate::phylo2vec::phylo2vec_quad;
     use crate::tree::Tree;
-    use crate::newick_to_vec;
+    use crate::newick::newick_to_vec;
     use crate::random_tree;
 
     #[test]