Include column 9 when writing complexes to GPI

Column 9: "Protein_Containing_Complex_Members"

Refs pombase/pombase-chado#1166

src/pombase/bio/go_format_writer.rs

@@ -162,10 +162,19 @@ fn write_complexes_to_gpi(gpi_writer: &mut dyn io::Write, protein_complex_map: &
 {
     let db_object_taxon = make_ncbi_taxon_id(config);
 
-    for protein_complex_uniquename in protein_complex_map.keys() {
-        let gpi_line = format!("ComplexPortal:{}\t\t\t\tGO:0032991\t{}\t\t\t\t\t\n",
+    for (protein_complex_uniquename, complex_details) in protein_complex_map.iter() {
+        let mut genes_with_prefixes: Vec<String> =
+            complex_details.genes
+               .iter().map(|gene_uniquename| {
+                format!("{}:{}", config.database_name, gene_uniquename)
+            }).collect();
+
+        genes_with_prefixes.sort();
+
+        let gpi_line = format!("ComplexPortal:{}\t\t\t\tGO:0032991\t{}\t\t\t{}\t\t\n",
                                protein_complex_uniquename,
-                               db_object_taxon);
+                               db_object_taxon,
+                               genes_with_prefixes.join("|"));
         gpi_writer.write_all(gpi_line.as_bytes())?;
     }
 

src/pombase/data_types.rs

@@ -2314,7 +2314,8 @@ pub struct ProteinViewData {
 
 #[derive(Serialize, Deserialize, Clone, Debug)]
 pub struct ProteinComplexDetails {
-    pub uniquename: FlexStr,
+    pub complex_uniquename: FlexStr,
+    pub genes: HashSet<GeneUniquename>,
 }
 
 pub type GoCamId = FlexStr;

src/pombase/web/data_build.rs

@@ -83,6 +83,7 @@ pub struct WebDataBuild<'a> {
     all_not_ont_annotations: HashMap<TermId, Vec<OntAnnotationId>>,
 
     protein_complexes: HashMap<ProteinComplexUniquename, ProteinComplexDetails>,
+    genes_of_complexes: HashMap<ProteinComplexUniquename, HashSet<GeneUniquename>>,
 
     // map from term name to term ID (ie "nucleus" -> "GO:0005634")
     term_ids_by_name: HashMap<FlexStr, TermId>,
@@ -879,6 +880,7 @@ impl <'a> WebDataBuild<'a> {
             transcripts: HashMap::new(),
             other_features: HashMap::new(),
             protein_complexes: HashMap::new(),
+            genes_of_complexes: HashMap::new(),
             terms: HashMap::new(),
             chromosomes: BTreeMap::new(),
             references: HashMap::new(),
@@ -1771,6 +1773,11 @@ phenotypes, so just the first part of this extension will be used:
                 entry.or_insert_with(Vec::new).push(part);
             }
 
+            if subject_type_name == "gene" && object_type_name == "protein-containing complex" {
+                let entry = self.genes_of_complexes.entry(object_uniquename.clone());
+                entry.or_insert_with(HashSet::new).insert(subject_uniquename.clone());
+            }
+
             if object_type_name == "genotype_interaction" {
                 let genotype_interaction_uniquename = object_uniquename.clone();
 
@@ -2237,7 +2244,9 @@ phenotypes, so just the first part of this extension will be used:
         let complex_uniquename = feat.uniquename.clone();
 
         let details = ProteinComplexDetails {
-            uniquename: complex_uniquename.clone()
+            complex_uniquename: complex_uniquename.clone(),
+            genes: self.genes_of_complexes.get(&complex_uniquename)
+              .map_or_else(HashSet::new, HashSet::clone),
         };
 
         self.protein_complexes.insert(complex_uniquename, details);