fix: more robust import of clinvar variants (#260)

varfish-org · Oct 18, 2023 · 72c8267 · 72c8267
1 parent 8318b94
commit 72c8267
Showing 1 changed file with 49 additions and 35 deletions.
diff --git a/src/clinvar_minimal/cli/import.rs b/src/clinvar_minimal/cli/import.rs
@@ -52,7 +52,13 @@ fn jsonl_import(
 
     for line in reader.lines() {
         let line = line?;
-        let record = serde_json::from_str::<clinvar_minimal::cli::reading::Record>(&line)?;
+        let record = match serde_json::from_str::<clinvar_minimal::cli::reading::Record>(&line) {
+            Ok(record) => record,
+            Err(e) => {
+                tracing::warn!("skipping line because of error: {}", e);
+                continue;
+            }
+        };
 
         let clinvar_minimal::cli::reading::Record {
             rcv,
@@ -85,42 +91,50 @@ fn jsonl_import(
             );
             let key: Vec<u8> = var.into();
 
-            let record = if let Some(data) = db
+            let data = db
                 .get_cf(&cf_data, key.clone())
-                .map_err(|e| anyhow::anyhow!("problem querying database: {}", e))?
-            {
-                let mut record = clinvar_minimal::pbs::Record::decode(&data[..])?;
-                record
-                    .reference_assertions
-                    .push(clinvar_minimal::pbs::ReferenceAssertion {
-                        rcv,
-                        title,
-                        clinical_significance: clinical_significance.into(),
-                        review_status: review_status.into(),
-                    });
-                record
-                    .reference_assertions
-                    .sort_by_key(|a| (a.clinical_significance, a.review_status));
-                record
-            } else {
-                clinvar_minimal::pbs::Record {
-                    release: assembly,
-                    chromosome: chr,
-                    start,
-                    stop,
-                    reference: reference_allele_vcf,
-                    alternative: alternate_allele_vcf,
-                    vcv,
-                    reference_assertions: vec![clinvar_minimal::pbs::ReferenceAssertion {
-                        rcv,
-                        title,
-                        clinical_significance: clinical_significance.into(),
-                        review_status: review_status.into(),
-                    }],
+                .map_err(|e| anyhow::anyhow!("problem querying database: {}", e));
+            match data {
+                Err(e) => {
+                    tracing::warn!("skipping line because of error: {}", e);
+                    continue;
+                }
+                Ok(data) => {
+                    let record = if let Some(data) = data {
+                        let mut record = clinvar_minimal::pbs::Record::decode(&data[..])?;
+                        record.reference_assertions.push(
+                            clinvar_minimal::pbs::ReferenceAssertion {
+                                rcv,
+                                title,
+                                clinical_significance: clinical_significance.into(),
+                                review_status: review_status.into(),
+                            },
+                        );
+                        record
+                            .reference_assertions
+                            .sort_by_key(|a| (a.clinical_significance, a.review_status));
+                        record
+                    } else {
+                        clinvar_minimal::pbs::Record {
+                            release: assembly,
+                            chromosome: chr,
+                            start,
+                            stop,
+                            reference: reference_allele_vcf,
+                            alternative: alternate_allele_vcf,
+                            vcv,
+                            reference_assertions: vec![clinvar_minimal::pbs::ReferenceAssertion {
+                                rcv,
+                                title,
+                                clinical_significance: clinical_significance.into(),
+                                review_status: review_status.into(),
+                            }],
+                        }
+                    };
+                    let buf = record.encode_to_vec();
+                    db.put_cf(&cf_data, key, buf)?;
                 }
-            };
-            let buf = record.encode_to_vec();
-            db.put_cf(&cf_data, key, buf)?;
+            }
         }
     }