feat: allow returning None for query interfaces (#255) (#256)

varfish-org · Oct 17, 2023 · b4bf349 · b4bf349
1 parent dd4eecc
commit b4bf349
Show file tree

Hide file tree

Showing 9 changed files with 191 additions and 108 deletions.
diff --git a/src/clinvar_genes/cli/query.rs b/src/clinvar_genes/cli/query.rs
@@ -74,6 +74,29 @@ fn print_record(
     Ok(())
 }
 
+/// Query for one gene annotation record.
+pub fn query_for_gene(
+    hgnc_id: &str,
+    db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
+    cf_data: &Arc<rocksdb::BoundColumnFamily>,
+) -> Result<Option<clinvar_genes::pbs::ClinvarPerGeneRecord>, anyhow::Error> {
+    let raw_value = db
+        .get_cf(cf_data, hgnc_id.as_bytes())
+        .map_err(|e| anyhow::anyhow!("error while querying for HGNC ID {}: {}", hgnc_id, e))?;
+    raw_value
+        .map(|raw_value| {
+            clinvar_genes::pbs::ClinvarPerGeneRecord::decode(&mut std::io::Cursor::new(&raw_value))
+                .map_err(|e| {
+                    anyhow::anyhow!(
+                        "error while decoding clinvar per gene record for HGNC ID {}: {}",
+                        hgnc_id,
+                        e
+                    )
+                })
+        })
+        .transpose()
+}
+
 /// Implementation of `gene query` sub command.
 pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error> {
     tracing::info!("Starting 'gene query' command");
@@ -94,19 +117,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     };
 
     tracing::info!("Running query...");
-    let raw_value = db.get_cf(&cf_data, args.hgnc_id.as_bytes())?;
-    if let Some(raw_value) = raw_value {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &clinvar_genes::pbs::ClinvarPerGeneRecord::decode(&mut std::io::Cursor::new(
-                &raw_value,
-            ))?,
-        )?;
+    if let Some(record) = query_for_gene(&args.hgnc_id, &db, &cf_data)? {
+        print_record(&mut out_writer, args.out_format, &record)?;
     } else {
-        tracing::info!("No data found for HGNC ID {}", args.hgnc_id);
+        tracing::info!("no record found for HGNC ID {:?}", args.hgnc_id);
     }
-
     tracing::info!("All done. Have a nice day!");
     Ok(())
 }
diff --git a/src/clinvar_minimal/cli/query.rs b/src/clinvar_minimal/cli/query.rs
@@ -102,7 +102,7 @@ pub fn query_for_variant(
     meta: &Meta,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     cf_data: &Arc<rocksdb::BoundColumnFamily>,
-) -> Result<pbs::Record, anyhow::Error> {
+) -> Result<Option<pbs::Record>, anyhow::Error> {
     // Split off the genome release (checked) and convert to key as used in database.
     let query = spdi::Var {
         sequence: extract_chrom::from_var(variant, Some(&meta.genome_release))?,
@@ -113,11 +113,15 @@ pub fn query_for_variant(
     let var: keys::Var = query.into();
     let key: Vec<u8> = var.into();
     let raw_value = db
-        .get_cf(cf_data, key)?
-        .ok_or_else(|| anyhow::anyhow!("could not find variant in database"))?;
-    // Decode via prost.
-    pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))
-        .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        .get_cf(cf_data, key)
+        .map_err(|e| anyhow::anyhow!("error while querying for variant {}: {}", variant, e))?;
+    raw_value
+        .map(|raw_value| {
+            // Decode via prost.
+            pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))
+                .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        })
+        .transpose()
 }
 
 /// Implementation of `tsv query` sub command.
@@ -141,11 +145,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     tracing::info!("Running query...");
     let before_query = std::time::Instant::now();
     if let Some(variant) = args.query.variant.as_ref() {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &query_for_variant(variant, &meta, &db, &cf_data)?,
-        )?;
+        if let Some(record) = query_for_variant(variant, &meta, &db, &cf_data)? {
+            print_record(&mut out_writer, args.out_format, &record)?;
+        } else {
+            tracing::info!("no record found for variant {:?}", &variant);
+        }
     } else {
         let (start, stop) = if let Some(position) = args.query.position.as_ref() {
             let position = spdi::Pos {

diff --git a/src/dbsnp/cli/query.rs b/src/dbsnp/cli/query.rs
@@ -118,7 +118,7 @@ pub fn query_for_variant(
     meta: &Meta,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     cf_data: &Arc<rocksdb::BoundColumnFamily>,
-) -> Result<dbsnp::pbs::Record, anyhow::Error> {
+) -> Result<Option<dbsnp::pbs::Record>, anyhow::Error> {
     // Split off the genome release (checked) and convert to key as used in database.
     let query = spdi::Var {
         sequence: extract_chrom::from_var(variant, Some(&meta.genome_release))?,
@@ -129,11 +129,15 @@ pub fn query_for_variant(
     let var: keys::Var = query.into();
     let key: Vec<u8> = var.into();
     let raw_value = db
-        .get_cf(cf_data, key)?
-        .ok_or_else(|| anyhow::anyhow!("could not find variant in database"))?;
-    // Decode via prost.
-    dbsnp::pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))
-        .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        .get_cf(cf_data, key)
+        .map_err(|e| anyhow::anyhow!("error while querying for variant {}: {}", variant, e))?;
+    raw_value
+        .map(|raw_value| {
+            // Decode via prost.
+            dbsnp::pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))
+                .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        })
+        .transpose()
 }
 
 /// Implementation of `tsv query` sub command.
@@ -157,11 +161,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     tracing::info!("Running query...");
     let before_query = std::time::Instant::now();
     if let Some(variant) = args.query.variant.as_ref() {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &query_for_variant(variant, &meta, &db, &cf_data)?,
-        )?;
+        if let Some(record) = query_for_variant(variant, &meta, &db, &cf_data)? {
+            print_record(&mut out_writer, args.out_format, &record)?;
+        } else {
+            tracing::info!("no record found for variant {}", variant);
+        }
     } else {
         let (start, stop) = if let Some(position) = args.query.position.as_ref() {
             let position = spdi::Pos {

diff --git a/src/freqs/cli/query.rs b/src/freqs/cli/query.rs
@@ -85,46 +85,58 @@ pub fn open_rocksdb_from_args(
     )
 }
 
+/// Enumeration of possible result records.
+pub enum Record {
+    /// Record for autosomal variant.
+    Autosomal(freqs::serialized::auto::Record),
+    /// Record for gonosomal variant.
+    Gonosomal(freqs::serialized::xy::Record),
+    /// Record for mitochondrial variant.
+    Mitochondrial(freqs::serialized::mt::Record),
+}
+
 /// Query for a single variant in the RocksDB database.
 pub fn query_for_variant(
     variant: &spdi::Var,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
-    out_writer: &mut dyn std::io::Write,
     _out_format: common::cli::OutputFormat,
-) -> Result<(), anyhow::Error> {
+) -> Result<Option<Record>, anyhow::Error> {
     let seq = variant.sequence.to_lowercase();
     let var: keys::Var = variant.clone().into();
     let key: Vec<u8> = var.into();
     if seq.contains('m') {
         let cf_mtdna: Arc<rocksdb::BoundColumnFamily> = db.cf_handle("mitochondrial").unwrap();
-        let raw_value = db.get_cf(&cf_mtdna, &key)?;
+        let raw_value = db
+            .get_cf(&cf_mtdna, &key)
+            .map_err(|e| anyhow::anyhow!("error reading from RocksDB: {}", e))?;
         if let Some(raw_value) = raw_value {
-            let value = freqs::serialized::mt::Record::from_buf(&raw_value);
-            let json_value = serde_json::to_value(value)?;
-            let json = serde_json::to_string(&json_value)?;
-            writeln!(out_writer, "{}", &json)?;
+            return Ok(Some(Record::Mitochondrial(
+                freqs::serialized::mt::Record::from_buf(&raw_value),
+            )));
         }
     } else if seq.contains('x') || seq.contains('y') {
         let cf_xy: Arc<rocksdb::BoundColumnFamily> = db.cf_handle("gonosomal").unwrap();
-        let raw_value = db.get_cf(&cf_xy, &key)?;
+        let raw_value = db
+            .get_cf(&cf_xy, &key)
+            .map_err(|e| anyhow::anyhow!("error reading from RocksDB: {}", e))?;
         if let Some(raw_value) = raw_value {
-            let value = freqs::serialized::xy::Record::from_buf(&raw_value);
-            let json_value = serde_json::to_value(value)?;
-            let json = serde_json::to_string(&json_value)?;
-            writeln!(out_writer, "{}", &json)?;
+            return Ok(Some(Record::Gonosomal(
+                freqs::serialized::xy::Record::from_buf(&raw_value),
+            )));
         }
     } else {
         let cf_auto: Arc<rocksdb::BoundColumnFamily> = db.cf_handle("autosomal").unwrap();
-        let raw_value = db.get_cf(&cf_auto, &key)?;
+        let raw_value = db
+            .get_cf(&cf_auto, &key)
+            .map_err(|e| anyhow::anyhow!("error reading from RocksDB: {}", e))?;
         if let Some(raw_value) = raw_value {
-            let value = freqs::serialized::auto::Record::from_buf(&raw_value);
-            let json_value = serde_json::to_value(value)?;
-            let json = serde_json::to_string(&json_value)?;
-            writeln!(out_writer, "{}", &json)?;
+            return Ok(Some(Record::Autosomal(
+                freqs::serialized::auto::Record::from_buf(&raw_value),
+            )));
         }
     }
 
-    Ok(())
+    Ok(None)
 }
 
 /// Implementation of `tsv query` sub command.
@@ -146,7 +158,27 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
 
     tracing::info!("Running query...");
     let before_query = std::time::Instant::now();
-    query_for_variant(&args.variant, &db, &mut out_writer, args.out_format)?;
+    if let Some(variant) = query_for_variant(&args.variant, &db, args.out_format)? {
+        match variant {
+            Record::Autosomal(record) => {
+                let json_value = serde_json::to_value(record)?;
+                let json = serde_json::to_string(&json_value)?;
+                writeln!(out_writer, "{}", &json)?;
+            }
+            Record::Gonosomal(record) => {
+                let json_value = serde_json::to_value(record)?;
+                let json = serde_json::to_string(&json_value)?;
+                writeln!(out_writer, "{}", &json)?;
+            }
+            Record::Mitochondrial(record) => {
+                let json_value = serde_json::to_value(record)?;
+                let json = serde_json::to_string(&json_value)?;
+                writeln!(out_writer, "{}", &json)?;
+            }
+        }
+    } else {
+        tracing::info!("no record found for variant {:?}", &args.variant);
+    }
     tracing::info!("... done querying in {:?}", before_query.elapsed());
 
     tracing::info!("All done. Have a nice day!");

diff --git a/src/genes/cli/query.rs b/src/genes/cli/query.rs
@@ -74,6 +74,28 @@ fn print_record(
     Ok(())
 }
 
+/// Query for one gene annotation record.
+pub fn query_for_gene(
+    hgnc_id: &str,
+    db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
+    cf_data: &Arc<rocksdb::BoundColumnFamily>,
+) -> Result<Option<pbs::Record>, anyhow::Error> {
+    let raw_value = db
+        .get_cf(cf_data, hgnc_id.as_bytes())
+        .map_err(|e| anyhow::anyhow!("error while querying for HGNC ID {}: {}", hgnc_id, e))?;
+    raw_value
+        .map(|raw_value| {
+            pbs::Record::decode(&mut std::io::Cursor::new(&raw_value)).map_err(|e| {
+                anyhow::anyhow!(
+                    "error while decoding gene record for HGNC ID {}: {}",
+                    hgnc_id,
+                    e
+                )
+            })
+        })
+        .transpose()
+}
+
 /// Implementation of `gene query` sub command.
 pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error> {
     tracing::info!("Starting 'gene query' command");
@@ -94,17 +116,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     };
 
     tracing::info!("Running query...");
-    let raw_value = db.get_cf(&cf_data, args.hgnc_id.as_bytes())?;
-    if let Some(raw_value) = raw_value {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))?,
-        )?;
+    if let Some(record) = query_for_gene(&args.hgnc_id, &db, &cf_data)? {
+        print_record(&mut out_writer, args.out_format, &record)?;
     } else {
-        tracing::info!("No data found for HGNC ID {}", args.hgnc_id);
+        tracing::info!("no record found for HGNC ID {:?}", args.hgnc_id);
     }
-
     tracing::info!("All done. Have a nice day!");
     Ok(())
 }
diff --git a/src/gnomad_mtdna/cli/query.rs b/src/gnomad_mtdna/cli/query.rs
@@ -102,7 +102,7 @@ pub fn query_for_variant(
     meta: &Meta,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     cf_data: &Arc<rocksdb::BoundColumnFamily>,
-) -> Result<gnomad_pbs::mtdna::Record, anyhow::Error> {
+) -> Result<Option<gnomad_pbs::mtdna::Record>, anyhow::Error> {
     // Split off the genome release (checked) and convert to key as used in database.
     let query = spdi::Var {
         sequence: extract_chrom::from_var(variant, Some(&meta.genome_release))?,
@@ -113,11 +113,15 @@ pub fn query_for_variant(
     let var: keys::Var = query.into();
     let key: Vec<u8> = var.into();
     let raw_value = db
-        .get_cf(cf_data, key)?
-        .ok_or_else(|| anyhow::anyhow!("could not find variant in database"))?;
-    // Decode via prost.
-    gnomad_pbs::mtdna::Record::decode(&mut std::io::Cursor::new(&raw_value))
-        .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        .get_cf(cf_data, key)
+        .map_err(|e| anyhow::anyhow!("error reading from RocksDB: {}", e))?;
+    raw_value
+        .map(|raw_value| {
+            // Decode via prost.
+            gnomad_pbs::mtdna::Record::decode(&mut std::io::Cursor::new(&raw_value))
+                .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        })
+        .transpose()
 }
 
 /// Implementation of `tsv query` sub command.
@@ -141,11 +145,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     tracing::info!("Running query...");
     let before_query = std::time::Instant::now();
     if let Some(variant) = args.query.variant.as_ref() {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &query_for_variant(variant, &meta, &db, &cf_data)?,
-        )?;
+        if let Some(record) = query_for_variant(variant, &meta, &db, &cf_data)? {
+            print_record(&mut out_writer, args.out_format, &record)?;
+        } else {
+            tracing::info!("no record found for variant {:?}", &variant);
+        }
     } else {
         let (start, stop) = if let Some(position) = args.query.position.as_ref() {
             let position = spdi::Pos {

diff --git a/src/gnomad_nuclear/cli/query.rs b/src/gnomad_nuclear/cli/query.rs
@@ -102,7 +102,7 @@ pub fn query_for_variant(
     meta: &Meta,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     cf_data: &Arc<rocksdb::BoundColumnFamily>,
-) -> Result<gnomad_pbs::gnomad2::Record, anyhow::Error> {
+) -> Result<Option<gnomad_pbs::gnomad2::Record>, anyhow::Error> {
     // Split off the genome release (checked) and convert to key as used in database.
     let query = spdi::Var {
         sequence: extract_chrom::from_var(variant, Some(&meta.genome_release))?,
@@ -113,11 +113,15 @@ pub fn query_for_variant(
     let var: keys::Var = query.into();
     let key: Vec<u8> = var.into();
     let raw_value = db
-        .get_cf(cf_data, key)?
-        .ok_or_else(|| anyhow::anyhow!("could not find variant in database"))?;
-    // Decode via prost.
-    gnomad_pbs::gnomad2::Record::decode(&mut std::io::Cursor::new(&raw_value))
-        .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        .get_cf(cf_data, key)
+        .map_err(|e| anyhow::anyhow!("problem querying RocksDB: {}", e))?;
+    raw_value
+        .map(|raw_value| {
+            // Decode via prost.
+            gnomad_pbs::gnomad2::Record::decode(&mut std::io::Cursor::new(&raw_value))
+                .map_err(|e| anyhow::anyhow!("failed to decode record: {}", e))
+        })
+        .transpose()
 }
 
 /// Implementation of `tsv query` sub command.
@@ -141,11 +145,11 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     tracing::info!("Running query...");
     let before_query = std::time::Instant::now();
     if let Some(variant) = args.query.variant.as_ref() {
-        print_record(
-            &mut out_writer,
-            args.out_format,
-            &query_for_variant(variant, &meta, &db, &cf_data)?,
-        )?;
+        if let Some(record) = query_for_variant(variant, &meta, &db, &cf_data)? {
+            print_record(&mut out_writer, args.out_format, &record)?
+        } else {
+            tracing::info!("no record found for variant {:?}", &variant);
+        }
     } else {
         let (start, stop) = if let Some(position) = args.query.position.as_ref() {
             let position = spdi::Pos {