fix: make "db mk-inhouse" work with BNS/INS (#322)

.github/workflows/release-please.yml

@@ -2,7 +2,7 @@ on:
   push:
     branches:
       - main
-      - releases/*
+      - release/*
 
 name: release-please
 

.github/workflows/rust.yml

@@ -4,11 +4,11 @@ on:
   push:
     branches:
       - main
-      - releases/*
+      - release/*
   pull_request:
     branches:
       - main
-      - releases/*
+      - release/*
 
 jobs:
   Formatting:

src/db/compile.rs

@@ -38,10 +38,10 @@ fn copy_db_def(
 
     debug!("Compute and verify MD5 checksum");
     // First create path to `.md5` file, then read in the MD5 string.
-    let md5_path = base_path.join(&format!("{file_name}{file_ext}.md5"));
-    let md5_str = read_md5_file(&md5_path)?.to_lowercase();
+    let md5_path = base_path.join(format!("{file_name}{file_ext}.md5"));
+    let md5_str = read_md5_file(md5_path)?.to_lowercase();
 
-    let file_path = base_path.join(&format!("{file_name}{file_ext}"));
+    let file_path = base_path.join(format!("{file_name}{file_ext}"));
     let chk_md5 = md5sum(&file_path)?.to_lowercase();
     if md5_str != chk_md5 {
         return Err(anyhow::anyhow!(

src/db/genes/data.rs

@@ -172,7 +172,7 @@ pub mod dbnsfp_gene {
     }
 
     /// Serialize `Vec<String>`, counterpart to `deserialize_vec`.
-    fn serialize_vec<S>(x: &Vec<String>, s: S) -> Result<S::Ok, S::Error>
+    fn serialize_vec<S>(x: &[String], s: S) -> Result<S::Ok, S::Error>
     where
         S: Serializer,
     {

src/db/mk_inhouse.rs

@@ -58,6 +58,8 @@ pub mod input {
         /// SV type of the record
         #[serde(deserialize_with = "from_varfish_sv_type")]
         pub sv_type: SvType,
+        /// Field with postgres-serialized JSON.
+        pub info: String,
         /// number of hom. alt. carriers
         pub num_hom_alt: u32,
         /// number of hom. ref. carriers
@@ -276,22 +278,29 @@ fn merge_to_out(
     // Read in all records and perform the "merge compression"
     let mut reader = JsonLinesReader::new(reader);
     while let Ok(Some(record)) = reader.read::<input::Record>() {
+        let info = record.info.replace("\"\"\"", "\"");
         let record = {
             let mut record = output::Record::from_db_record(record);
             record.begin -= 1; // need to get 0-based coordinates for DB
+            record.end = match record.sv_type {
+                SvType::Bnd => {
+                    // Obtain "pos2" from JSON-encoded info field.
+                    let info = serde_json::from_str::<serde_json::Value>(&info)?;
+                    info.as_object()
+                        .and_then(|o| o.get("pos2"))
+                        .and_then(|v| v.as_i64())
+                        .ok_or_else(|| anyhow::anyhow!("Cannot find 'pos2' in info field"))?
+                        as i32
+                }
+                SvType::Ins => record.begin + 1,
+                _ => record.end,
+            };
             record
         };
-        let begin = match record.sv_type {
-            SvType::Bnd => record.begin - 1 - args.slack_bnd,
-            SvType::Ins => record.begin - 1 - args.slack_ins,
-            _ => record.begin,
+        let query = match record.sv_type {
+            SvType::Bnd => record.begin..(record.begin + 1),
+            _ => record.begin..record.end,
         };
-        let end = match record.sv_type {
-            SvType::Bnd => record.begin + args.slack_bnd,
-            SvType::Ins => record.begin + args.slack_ins,
-            _ => record.end,
-        };
-        let query = begin..end;
         let mut found_any_cluster = false;
         for mut it_tree in tree.find_mut(&query) {
             let cluster_idx = *it_tree.data();
@@ -317,13 +326,7 @@ fn merge_to_out(
         }
         if !found_any_cluster {
             // create new cluster
-            tree.insert(
-                match record.sv_type {
-                    SvType::Bnd | SvType::Ins => (record.begin - 1)..record.begin,
-                    _ => (record.begin - 1)..record.end,
-                },
-                clusters.len(),
-            );
+            tree.insert(query, clusters.len());
             clusters.push(vec![records.len()]);
         }
         // always register the record

src/main.rs

@@ -142,11 +142,6 @@ fn main() -> Result<(), anyhow::Error> {
         .compact()
         .finish();
 
-    // Common variables for CLI commands from annonars.
-    let annonars_common = annonars::common::cli::Args {
-        verbose: cli.common.verbose.clone(),
-    };
-
     // Install collector and go into sub commands.
     let term = Term::stderr();
     tracing::subscriber::with_default(collector, || {

src/pheno/algos.rs

@@ -70,11 +70,10 @@ mod test {
             .omim_disease(&OmimDiseaseId::from(154700))
             .expect("marfan symdrome must be in HPO");
         let hpo_marfan = HpoGroup::from_iter(
-            omim_marfan
+            &omim_marfan
                 .to_hpo_set(&hpo)
                 .child_nodes()
-                .without_modifier()
-                .into_iter(),
+                .without_modifier(),
         );
 
         let score = phenomizer::score(&prepare(query), &hpo_marfan, &hpo);

src/pheno/prepare.rs

@@ -100,7 +100,7 @@ fn run_simulation(
                 let path = std::path::Path::new(path_gene_logs).join(num_terms.to_string());
                 std::fs::create_dir_all(&path).expect("cannot create logs directory");
                 Some(
-                    std::fs::File::create(&format!("{}/{}.txt", path.display(), gene.symbol()))
+                    std::fs::File::create(format!("{}/{}.txt", path.display(), gene.symbol()))
                         .expect("could not open file"),
                 )
             } else {
@@ -130,10 +130,7 @@ fn run_simulation(
                     let s = phenomizer::score(
                         &ts,
                         &HpoGroup::from_iter(
-                            gene.to_hpo_set(ontology)
-                                .child_nodes()
-                                .without_modifier()
-                                .into_iter(),
+                            &gene.to_hpo_set(ontology).child_nodes().without_modifier(),
                         ),
                         ontology,
                     );
@@ -147,11 +144,7 @@ fn run_simulation(
                             s,
                             gene.symbol(),
                             ts.iter()
-                                .map(|t| format!(
-                                    "{} ({})",
-                                    t.to_string(),
-                                    ontology.hpo(t).unwrap().name()
-                                ))
+                                .map(|t| format!("{} ({})", t, ontology.hpo(t).unwrap().name()))
                                 .collect::<Vec<_>>()
                                 .join(", ")
                         )

src/pheno/query.rs

@@ -248,7 +248,7 @@ pub fn run(args_common: &crate::common::Args, args: &Args) -> Result<(), anyhow:
     let path_rocksdb = format!("{}/resnik", args.path_hpo_dir);
     let db = rocksdb::DB::open_cf_for_read_only(
         &rocksdb::Options::default(),
-        &path_rocksdb,
+        path_rocksdb,
         ["meta", "resnik_pvalues"],
         true,
     )?;

src/server/annos.rs

@@ -552,6 +552,7 @@ pub mod actix_server {
         #[serde_with::serde_as]
         #[derive(Deserialize, Debug, Clone)]
         #[serde(rename_all = "kebab-case")]
+        #[allow(dead_code)]
         struct Request {
             /// Genome release version.
             pub genome_release: String,
@@ -730,7 +731,7 @@ pub fn run(args_common: &crate::common::Args, args: &Args) -> Result<(), anyhow:
     tracing::info!("Opening databases...");
     let before_opening = Instant::now();
     // Argument lists from the command line with the corresponding database enum value.
-    let paths_db_pairs = vec![
+    let paths_db_pairs = [
         (&args.path_cadd, AnnoDb::Cadd),
         (&args.path_dbnsfp, AnnoDb::Dbnsfp),
         (&args.path_dbsnp, AnnoDb::Dbsnp),

src/server/pheno.rs

@@ -494,20 +494,18 @@ pub mod actix_server {
             ) -> Self {
                 let hpo_terms = if hpo_terms {
                     Some(
-                        HpoGroup::from_iter(
-                            omim_disease.to_hpo_set(ontology).child_nodes().into_iter(),
-                        )
-                        .into_iter()
-                        .map(|term_id| ontology.hpo(term_id))
-                        .filter(|term| term.is_some())
-                        .map(|term| {
-                            let term = term.expect("filtered above");
-                            ResultHpoTerm {
-                                term_id: term.id().to_string(),
-                                name: term.name().to_string(),
-                            }
-                        })
-                        .collect(),
+                        HpoGroup::from_iter(&omim_disease.to_hpo_set(ontology).child_nodes())
+                            .into_iter()
+                            .map(|term_id| ontology.hpo(term_id))
+                            .filter(|term| term.is_some())
+                            .map(|term| {
+                                let term = term.expect("filtered above");
+                                ResultHpoTerm {
+                                    term_id: term.id().to_string(),
+                                    name: term.name().to_string(),
+                                }
+                            })
+                            .collect(),
                     )
                 } else {
                     None

src/server/rest.rs

@@ -356,7 +356,7 @@ pub fn run(args_common: &crate::common::Args, args: &Args) -> Result<(), anyhow:
         } else {
             PathBuf::from_str(&args.path_db)?.join("conf.toml")
         };
-        let toml_str = std::fs::read_to_string(&path_conf)?;
+        let toml_str = std::fs::read_to_string(path_conf)?;
         toml::from_str(&toml_str)?
     };