Add QueryResultFilter module for filtering query results; Set lookup …

…sep char to "\t"

src/controller/filter.rs

@@ -0,0 +1,50 @@
+// Result Filtering module
+use crate::controller::rank::QueryResult;
+
+// NOTE: Limiting result with top N results is done in CLI
+// This module is for filtering single query result
+
+pub struct QueryResultFilter {
+    // Filtering parameters that doesn't require residue matching
+    pub total_hash_match_count: usize,
+    pub total_hash_match_ratio: f32,
+    pub node_covered_by_hash_count: usize,
+    pub node_covered_by_hash_ratio: f32,
+    pub edge_covered_by_hash_count: usize,
+    pub edge_covered_by_hash_ratio: f32,
+    pub idf: f32,
+    pub nres: usize,
+    pub plddt: f32,
+    // Filtering parameters that require residue matching
+    pub node_covered_by_graph_count: usize,
+    pub node_covered_by_graph_ratio: f32,
+    pub rmsd: f32,
+}
+
+impl QueryResultFilter {
+    pub fn new() -> Self {
+        QueryResultFilter {
+            total_hash_match_count: 0,
+            total_hash_match_ratio: 0.0,
+            node_covered_by_hash_count: 0,
+            node_covered_by_hash_ratio: 0.0,
+            edge_covered_by_hash_count: 0,
+            edge_covered_by_hash_ratio: 0.0,
+            idf: 0.0,
+            nres: 0,
+            plddt: 0.0,
+            node_covered_by_graph_count: 0,
+            node_covered_by_graph_ratio: 0.0,
+            rmsd: 0.0,
+        }
+    }
+    // Filter single query result
+    pub fn filter(&self, result: &QueryResult) -> bool {
+        // let mut pass = true;
+        // if self.total_hash_match_count > 0 {
+        //     pass = pass && result.total_hash_match_count >= self.total_hash_match_count;
+        // }
+        todo!()
+    }
+
+}

src/controller/mod.rs

@@ -6,6 +6,7 @@
 // Copyright © 2024 Hyunbin Kim, All rights reserved
 
 pub mod feature;
+// pub mod filter;
 pub mod graph;
 pub mod io;
 pub mod query;

src/index/lookup.rs

@@ -69,7 +69,7 @@ pub fn load_lookup_from_file(path: &str) -> Vec<(String, usize, usize, f32)> {
     let mmap = unsafe { Mmap::map(&file).expect(&log_msg(FAIL, "Unable to mmap the lookup file")) };
     let content = unsafe { std::str::from_utf8_unchecked(&mmap) };
     let loaded_lookup = content.par_lines().map(|line| {
-        let mut split = line.split_whitespace();
+        let mut split = line.split("\t");
         let id = split.next().unwrap().parse::<usize>().unwrap();
         let name = split.next().unwrap().to_string();
         let nres = split.next().unwrap().parse::<usize>().unwrap();

src/structure/io/fcz.rs

@@ -167,7 +167,7 @@ pub fn read_foldcomp_db_lookup(db_path: &str) -> Result<Vec<(usize, String)>, &'
     let mmap = unsafe { Mmap::map(&lookup_file).unwrap() };
     let content = unsafe { std::str::from_utf8_unchecked(&mmap) };
     let output = content.par_lines().map(|line| {
-        let mut split = line.split_whitespace();
+        let mut split = line.split("\t");
         let id = split.next().unwrap().parse::<usize>().unwrap();
         let name = split.next().unwrap().to_string();
         (id, name)
@@ -186,7 +186,7 @@ pub fn read_foldcomp_db_index(db_path: &str) -> Result<Vec<(usize, usize, usize)
     let mmap = unsafe { Mmap::map(&index_file).unwrap() };
     let content = unsafe { std::str::from_utf8_unchecked(&mmap) };
     let output = content.par_lines().map(|line| {
-        let mut split = line.split_whitespace();
+        let mut split = line.split("\t");
         let id = split.next().unwrap().parse::<usize>().unwrap();
         let start = split.next().unwrap().parse::<usize>().unwrap();
         let length = split.next().unwrap().parse::<usize>().unwrap();