Filter crap proteins after analyzing the peptides

Author: tibvdm

api/src/controllers/mpa/mod.rs

@@ -11,3 +11,7 @@ pub fn default_tryptic() -> bool { false }
 pub fn default_report_taxa() -> bool {
     false
 }
+
+pub fn default_blacklist_crap() -> bool {
+    false
+}

api/src/controllers/mpa/pept2data.rs

@@ -3,14 +3,15 @@ use axum::{extract::State, Json};
 use serde::{Deserialize, Serialize};
 use index::{ProteinInfo, SearchResult};
 use crate::{
-    controllers::{generate_handlers, mpa::default_equate_il, mpa::default_tryptic, mpa::default_report_taxa, api::default_cutoff},
+    controllers::{generate_handlers, mpa::default_equate_il, mpa::default_tryptic, mpa::default_report_taxa, mpa::default_blacklist_crap, api::default_cutoff},
     helpers::{
         fa_helper::{calculate_fa, FunctionalAggregation},
         lca_helper::calculate_lca,
         lineage_helper::{get_lineage_array, LineageVersion}
     },
     AppState
 };
+use crate::helpers::filters::crap_filter::CrapFilter;
 use crate::helpers::filters::empty_filter::EmptyFilter;
 use crate::helpers::filters::protein_filter::ProteinFilter;
 use crate::helpers::filters::proteome_filter::ProteomeFilter;
@@ -30,6 +31,8 @@ pub struct Parameters {
     cutoff: usize,
     #[serde(default = "default_report_taxa")]
     report_taxa: bool,
+    #[serde(default = "default_blacklist_crap")]
+    blacklist_crap: bool,
     filter: Option<Filter>,
 }
 
@@ -60,7 +63,7 @@ pub struct Data {
 
 async fn handler(
     State(AppState { index, datastore, .. }): State<AppState>,
-    Parameters { mut peptides, equate_il, tryptic, cutoff, report_taxa, filter }: Parameters
+    Parameters { mut peptides, equate_il, tryptic, cutoff, report_taxa, blacklist_crap, filter }: Parameters
 ) -> Result<Data, ()> {
     if peptides.is_empty() {
         return Ok(Data { peptides: Vec::new() });
@@ -93,13 +96,16 @@ async fn handler(
         None => Box::new(EmptyFilter::new())
     };
 
+    let crap_blacklist = CrapFilter::new();
+
     Ok(Data {
         peptides: result
             .into_iter()
             .filter_map(|SearchResult { proteins, sequence, .. }| {
                 let filtered_proteins: Vec<ProteinInfo> = proteins
                     .into_iter()
                     .filter(|protein| filter_proteins.filter(protein))
+                    .filter(|protein| !(blacklist_crap && crap_blacklist.filter(protein)))
                     .collect();
 
                 if filtered_proteins.is_empty() {

api/src/helpers/filters/crap_filter.rs

@@ -0,0 +1,49 @@
+use std::collections::HashSet;
+use index::ProteinInfo;
+use crate::helpers::filters::protein_filter::ProteinFilter;
+use crate::helpers::filters::UniprotFilter;
+
+pub struct CrapFilter {
+    protein_filter: ProteinFilter,
+}
+
+impl UniprotFilter for CrapFilter {
+    fn filter(&self, protein: &ProteinInfo) -> bool {
+        self.protein_filter.filter(protein)
+    }
+}
+
+impl CrapFilter {
+    pub fn new() -> Self {
+        CrapFilter {
+            protein_filter: ProteinFilter::new(Self::get_crap_proteins()),
+        }
+    }
+
+    fn get_crap_proteins() -> HashSet<String> {
+        let crap_accessions = vec![
+            "P02769", "P0DTE7", "P0DTE8", "P0DUB6", "P02662", "P02663",
+            "P02666", "P02668", "P00766", "P00767", "P13645", "O77727",
+            "P35527", "Q15323", "Q14532", "O76011", "Q92764", "O76013",
+            "O76014", "O76015", "O76009", "P01920", "P02534", "P02539",
+            "P35908", "P04264", "P02539", "P15241", "P25691", "P02444",
+            "P02445", "P02443", "P02441", "Q02958", "P02438", "P02439",
+            "P02440", "P08131", "Q14533", "Q9NSB4", "P78385", "Q9NSB2",
+            "P78386", "O43790", "P26372", "P00711", "Q7M135", "P00792",
+            "P00791", "Q10735", "P30879", "P0C1U8", "P00760", "Q29463",
+            "A0A8K0BFD9", "P02768", "P01008", "D6RCN3", "P61769", "P55957",
+            "P00915", "P00918", "P04040", "P07339", "P08311", "P01031",
+            "P02741", "P00167", "P99999", "P01133", "P05413", "P06396",
+            "Q9BX51", "A0A2R8Y5E5", "P69905", "P68871", "P01344", "P10145",
+            "P06732", "P00709", "P80384", "P61626", "P02144", "Q15843",
+            "P15559", "U3KQG7", "P01127", "P62937", "A0A0A0MRQ5", "P01112",
+            "P02753", "P00441", "B8ZZN6", "P12081", "P10636", "P10599",
+            "P01375", "P02787", "P02788", "P51965", "O00762", "A8MUA9",
+            "P62979", "P32503", "P00004", "P00921", "P00330", "P00883",
+            "P00698", "P68082", "P01012", "P00722", "P00366", "A0A5J6CYK8",
+            "A0A182BM84", "P15252", "P81054"
+        ];
+
+        crap_accessions.into_iter().map(String::from).collect()
+    }
+}

api/src/helpers/filters/mod.rs

@@ -4,6 +4,7 @@ pub mod taxa_filter;
 pub mod proteome_filter;
 pub mod protein_filter;
 pub mod empty_filter;
+pub mod crap_filter;
 
 pub trait UniprotFilter {
     fn filter(&self, protein: &ProteinInfo) -> bool;