Reorganized modules

src/bin/ringo_search.rs

@@ -4,7 +4,12 @@ use ringo::db::search::similarity_search;
 
 fn main() {
     let args: Vec<String> = std::env::args().collect();
-    let results = similarity_search(&args[1], &args[2], args[3].parse().unwrap(), args[4].parse().unwrap());
+    let results = similarity_search(
+        &args[1],
+        &args[2],
+        args[3].parse().unwrap(),
+        args[4].parse().unwrap(),
+    );
     for result in results {
         println!("{:?} {:?}", result.line, result.similarity);
     }

src/db.rs

@@ -1,4 +1,3 @@
-pub mod index_item;
 pub mod index;
+pub mod index_item;
 pub mod search;
-pub mod fingerprint;

src/db/index.rs

@@ -1,10 +1,9 @@
-
+use crate::db::index_item::IndexItem;
+use crate::io::smiles::reader::molecule::parse_molecule;
+use crate::model::fingerprint::FINGERPRINT_SIZE;
 use bincode::encode_into_slice;
 use std::fs::File;
 use std::io::{BufRead, BufWriter, Write};
-use crate::db::index_item::IndexItem;
-use crate::db::fingerprint::FINGERPRINT_SIZE;
-use crate::molecule::smiles::reader::molecule::parse_molecule;
 
 #[cfg(windows)]
 const LINE_ENDING_LENGTH: usize = 2;
@@ -36,7 +35,6 @@ pub fn index_file(smiles_file: &str) {
     }
 }
 
-
 #[cfg(test)]
 mod test {
     use crate::db::index::index_file;

src/db/index_item.rs

@@ -1,4 +1,4 @@
-use crate::db::fingerprint::Fingerprint;
+use crate::model::fingerprint::Fingerprint;
 use bincode::{Decode, Encode};
 
 #[derive(Debug, Encode, Decode)]
@@ -9,11 +9,11 @@ pub struct IndexItem {
 
 #[cfg(test)]
 mod tests {
+    use crate::db::index_item::IndexItem;
+    use crate::model::fingerprint::Fingerprint;
     use bincode::config::standard;
     use bincode::{decode_from_slice, encode_to_vec};
     use fixedbitset::FixedBitSet;
-    use crate::db::fingerprint::Fingerprint;
-    use crate::db::index_item::IndexItem;
 
     #[test]
     fn test_index_item_encode_decode() {

src/db/search.rs

@@ -1,6 +1,6 @@
-use crate::math::similarity::tanimoto::tanimoto_bitset;
-use crate::molecule::smiles::reader::molecule::parse_molecule;
 use crate::db::index_item::IndexItem;
+use crate::io::smiles::reader::molecule::parse_molecule;
+use crate::math::similarity::tanimoto::tanimoto_bitset;
 use std::fs::File;
 use std::io::{BufRead, BufReader, Read, Seek};
 
@@ -67,10 +67,6 @@ pub fn similarity_search(
     results
 }
 
-fn main() {
-    println!("db-search");
-}
-
 #[cfg(test)]
 mod test {
     use crate::db::index::index_file;

src/descriptors.rs

@@ -0,0 +1,2 @@
+mod fingerprints;
+mod weight;

src/descriptors/fingerprints.rs

@@ -0,0 +1 @@
+mod ecfp;

src/descriptors/fingerprints/ecfp.rs

@@ -0,0 +1,84 @@
+use crate::model::atom::Atom;
+use crate::model::bond::Bond;
+use crate::model::fingerprint::{Fingerprint, FINGERPRINT_SIZE};
+use crate::model::molecule::Molecule;
+use fixedbitset::FixedBitSet;
+use petgraph::graph::NodeIndex;
+use petgraph::prelude::{EdgeRef, StableGraph};
+use petgraph::Undirected;
+use std::hash::{DefaultHasher, Hasher};
+
+impl Molecule {
+    pub fn ecfp(&self, radius: usize, fp_length: usize) -> Fingerprint {
+        let mut fp = FixedBitSet::with_capacity(FINGERPRINT_SIZE);
+
+        for node in self.graph.node_indices() {
+            ecfp_recursive(
+                &self.graph,
+                radius,
+                1,
+                node,
+                &mut fp,
+                fp_length,
+                &mut DefaultHasher::new(),
+            );
+        }
+
+        Fingerprint(fp)
+    }
+}
+
+fn ecfp_recursive(
+    graph: &StableGraph<Atom, Bond, Undirected>,
+    radius: usize,
+    depth: usize,
+    node: NodeIndex,
+    fp: &mut FixedBitSet,
+    fp_length: usize,
+    hasher: &mut DefaultHasher,
+) {
+    if depth > radius {
+        return;
+    }
+
+    let atom = graph.node_weight(node).unwrap();
+    hasher.write_u8(atom.element.atomic_number);
+    hasher.write_u8(atom.isotope);
+    hasher.write_i8(atom.charge);
+    hasher.write_u8(atom.hs);
+
+    let value = hasher.clone().finish();
+    fp.insert(value as usize % fp_length);
+
+    for edge in graph.edges(node) {
+        let bond = edge.weight();
+        hasher.write_u8(bond.order as u8);
+
+        let target = if edge.source() == node {
+            edge.target()
+        } else {
+            edge.source()
+        };
+
+        ecfp_recursive(graph, radius, depth + 1, target, fp, fp_length, hasher);
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use crate::io::smiles::reader::molecule::parse_molecule;
+    use crate::math::similarity::tanimoto::tanimoto_bitset;
+    #[test]
+    fn test_ecfp() {
+        let ecfp_ibuprofen = parse_molecule("CC(C)CC1=CC=C(C=C1)C(C)C(=O)O")
+            .unwrap()
+            .1
+            .ecfp(2, 128);
+        let ecfp_naproxen = parse_molecule("CC(C1=CC2=C(C=C1)C=C(C=C2)OC)C(=O)O")
+            .unwrap()
+            .1
+            .ecfp(2, 128);
+        let sim = tanimoto_bitset(&ecfp_ibuprofen.0, &ecfp_naproxen.0);
+        assert!(0.53 < sim && sim < 0.54);
+    }
+}

src/descriptors/weight.rs

@@ -0,0 +1,11 @@
+use crate::model::molecule::Molecule;
+
+impl Molecule {
+    pub fn weight(&self) -> f64 {
+        let mut weight: f64 = 0.0;
+        for atom in self.graph.node_weights() {
+            weight += atom.element.atomic_weight();
+        }
+        weight
+    }
+}

src/io.rs

@@ -0,0 +1 @@
+pub(crate) mod smiles;

src/molecule/smiles/reader/atom.rs → src/io/smiles/reader/atom.rs

@@ -1,9 +1,9 @@
-use crate::molecule::model::atom::Atom;
-use crate::molecule::model::element::Element;
-use crate::molecule::smiles::reader::charge::parse_charge;
-use crate::molecule::smiles::reader::element::parse_element;
-use crate::molecule::smiles::reader::hydrogens::parse_hydrogens;
-use crate::molecule::smiles::reader::isotope::parse_isotope;
+use crate::io::smiles::reader::charge::parse_charge;
+use crate::io::smiles::reader::element::parse_element;
+use crate::io::smiles::reader::hydrogens::parse_hydrogens;
+use crate::io::smiles::reader::isotope::parse_isotope;
+use crate::model::atom::Atom;
+use crate::model::element::Element;
 use nom::combinator::opt;
 use nom::IResult;
 
@@ -47,7 +47,7 @@ pub fn parse_atom(input: &str) -> IResult<&str, Atom> {
 
 #[cfg(test)]
 mod tests {
-    use crate::molecule::smiles::reader::atom::parse_atom;
+    use crate::io::smiles::reader::atom::parse_atom;
 
     fn do_test_parse_atom(input: &str, atomic_number: u8, charge: i8, hs: u8, isotope: u8) {
         let (remaining_input, atom) = parse_atom(input).unwrap();

src/molecule/smiles/reader/bond.rs → src/io/smiles/reader/bond.rs

@@ -1,4 +1,4 @@
-use crate::molecule::model::bond::{Bond, BondOrder};
+use crate::model::bond::{Bond, BondOrder};
 use nom::branch::alt;
 use nom::bytes::complete::tag;
 use nom::combinator::map;
@@ -16,8 +16,8 @@ pub fn parse_bond(input: &str) -> IResult<&str, Bond> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::model::bond::BondOrder;
-    use crate::ringo::molecule::smiles::reader::bond::parse_bond;
+    use crate::io::smiles::reader::bond::parse_bond;
+    use crate::model::bond::BondOrder;
 
     #[test]
     fn parse_bond_empty() {

src/molecule/smiles/reader/charge.rs → src/io/smiles/reader/charge.rs

@@ -33,7 +33,7 @@ pub fn parse_charge(input: &str) -> IResult<&str, i8> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::smiles::reader::charge::parse_charge;
+    use crate::io::smiles::reader::charge::parse_charge;
     use nom::error::{Error, ErrorKind};
 
     #[test]

src/molecule/smiles/reader/element.rs → src/io/smiles/reader/element.rs

@@ -35,7 +35,7 @@ pub fn parse_element(input: &str) -> IResult<&str, u8> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::smiles::reader::element::parse_element;
+    use crate::io::smiles::reader::element::parse_element;
     use nom::error::{Error, ErrorKind};
 
     #[test]

src/molecule/smiles/reader/hydrogens.rs → src/io/smiles/reader/hydrogens.rs

@@ -28,7 +28,7 @@ pub fn parse_hydrogens(input: &str) -> IResult<&str, u8> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::smiles::reader::hydrogens::parse_hydrogens;
+    use crate::io::smiles::reader::hydrogens::parse_hydrogens;
     use nom::error::{Error, ErrorKind};
 
     #[test]

src/molecule/smiles/reader/isotope.rs → src/io/smiles/reader/isotope.rs

@@ -8,7 +8,7 @@ pub fn parse_isotope(input: &str) -> IResult<&str, u8> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::smiles::reader::isotope::parse_isotope;
+    use crate::io::smiles::reader::isotope::parse_isotope;
     use nom::error::{Error, ErrorKind};
 
     #[test]

src/molecule/smiles/reader/molecule.rs → src/io/smiles/reader/molecule.rs

@@ -1,7 +1,7 @@
-use crate::molecule::model::bond::{Bond, BondOrder};
-use crate::molecule::model::molecule::Molecule;
-use crate::molecule::smiles::reader::atom::parse_atom;
-use crate::molecule::smiles::reader::bond::parse_bond;
+use crate::io::smiles::reader::atom::parse_atom;
+use crate::io::smiles::reader::bond::parse_bond;
+use crate::model::bond::{Bond, BondOrder};
+use crate::model::molecule::Molecule;
 use nom::branch::alt;
 use nom::character::complete::{char, digit1};
 use nom::combinator::{map, map_res};
@@ -90,8 +90,8 @@ pub fn parse_molecule(input: &str) -> IResult<&str, Molecule> {
 
 #[cfg(test)]
 mod tests {
-    use crate::ringo::molecule::model::bond::BondOrder;
-    use crate::ringo::molecule::smiles::reader::molecule::parse_molecule;
+    use crate::io::smiles::reader::molecule::parse_molecule;
+    use crate::model::bond::BondOrder;
     use petgraph::stable_graph::{EdgeIndex, NodeIndex};
 
     #[test]

src/lib.rs

@@ -1,3 +1,5 @@
 pub mod db;
+mod descriptors;
+mod io;
 mod math;
-mod molecule;
+mod model;

src/molecule/model.rs → src/model.rs

@@ -1,4 +1,5 @@
 pub mod atom;
 pub mod bond;
 pub mod element;
+pub mod fingerprint;
 pub mod molecule;

src/molecule/model/atom.rs → src/model/atom.rs

@@ -1,4 +1,4 @@
-use crate::molecule::model::element::Element;
+use crate::model::element::Element;
 
 #[derive(Hash, Eq, PartialEq, Debug)]
 pub struct Atom {

src/molecule/model/element.rs → src/model/element.rs

@@ -96,7 +96,7 @@ impl Element {
 
 #[cfg(test)]
 mod test {
-    use crate::ringo::molecule::model::element::Element;
+    use crate::model::element::Element;
 
     #[test]
     fn test_element() {

src/db/fingerprint.rs → src/model/fingerprint.rs

@@ -32,7 +32,7 @@ impl<'de> bincode::BorrowDecode<'de> for Fingerprint {
 
 #[cfg(test)]
 mod tests {
-    use crate::db::fingerprint::{Fingerprint, FINGERPRINT_SIZE};
+    use crate::model::fingerprint::{Fingerprint, FINGERPRINT_SIZE};
     use fixedbitset::FixedBitSet;
 
     #[test]