speedups

rust/bioscript-cli/src/report_html_observations.rs

@@ -41,6 +41,7 @@ fn render_observation_table(
     let show_facets = observations
         .iter()
         .any(|observation| !json_field_as_tsv(observation.get("facets")).is_empty());
+    let show_imputed_reference_note = observations.iter().any(observation_is_imputed_vcf_reference);
     let headers = all_headers
         .iter()
         .copied()
@@ -66,6 +67,9 @@ fn render_observation_table(
         out.push_str("</tr>");
     }
     out.push_str("</tbody></table></div>");
+    if show_imputed_reference_note {
+        out.push_str("<p class=\"muted observation-note\">* In variant-only VCF inputs, absent queried variant rows are shown as imputed reference genotypes. This is usually appropriate for variant-only VCFs, but it may be wrong if the VCF omits loci for another reason.</p>");
+    }
 }
 
 fn observation_filter_group(observation: &serde_json::Value) -> &'static str {
@@ -125,6 +129,14 @@ fn observation_row_class(observation: &serde_json::Value) -> &'static str {
 
 fn render_observation_cell(out: &mut String, observation: &serde_json::Value, header: &str) {
     let cell_class = table_column_class(header);
+    if header == "outcome" {
+        let mut value = json_field_as_tsv(observation.get(header));
+        if value == "reference" && observation_is_imputed_vcf_reference(observation) {
+            value.push('*');
+        }
+        let _ = write!(out, "<td class=\"{}\">{}</td>", cell_class, html_escape(&value));
+        return;
+    }
     if header == "ref_alt" {
         class_cell(out, &observation_ref_alt(observation), "mono");
         return;
@@ -184,6 +196,15 @@ fn render_observation_cell(out: &mut String, observation: &serde_json::Value, he
     );
 }
 
+fn observation_is_imputed_vcf_reference(observation: &serde_json::Value) -> bool {
+    observation
+        .get("evidence_raw")
+        .and_then(serde_json::Value::as_str)
+        .is_some_and(|evidence| {
+            evidence.contains("imputed reference genotype from absent variant-only VCF record")
+        })
+}
+
 fn observation_ref_alt(observation: &serde_json::Value) -> String {
     let ref_allele = observation
         .get("ref")

rust/bioscript-cli/src/report_html_sections.rs

@@ -168,6 +168,7 @@ fn render_input_debug(out: &mut String, reports: &[serde_json::Value], show_part
         "Source",
         "Assembly",
         "Inferred Sex",
+        "VCF Ref Imputation",
         "Evidence",
     ]);
     for (idx, header) in headers.iter().enumerate() {
@@ -217,6 +218,7 @@ fn render_input_debug(out: &mut String, reports: &[serde_json::Value], show_part
                 value_str(sex, "method"),
             ]),
         );
+        table_cell(out, input_debug_vcf_imputation(debug));
         table_cell(out, &input_debug_evidence(debug));
         out.push_str("</tr>");
     }
@@ -256,6 +258,7 @@ fn render_input_debug_key_values(out: &mut String, report: &serde_json::Value) {
             value_str(sex, "method"),
         ]),
     );
+    input_debug_kv(out, "VCF ref imputation", input_debug_vcf_imputation(debug));
     input_debug_kv(out, "Evidence", &input_debug_evidence(debug));
     out.push_str("</dl>");
 }
@@ -278,6 +281,18 @@ fn compact_join(values: &[&str]) -> String {
         .join(" / ")
 }
 
+fn input_debug_vcf_imputation(debug: &serde_json::Value) -> &'static str {
+    if debug
+        .get("vcf_missing_reference_imputation")
+        .and_then(serde_json::Value::as_bool)
+        == Some(true)
+    {
+        "used"
+    } else {
+        ""
+    }
+}
+
 fn input_debug_evidence(debug: &serde_json::Value) -> String {
     let mut evidence = Vec::new();
     collect_string_array(debug.get("evidence"), &mut evidence);
@@ -300,4 +315,3 @@ fn collect_string_array(value: Option<&serde_json::Value>, out: &mut Vec<String>
         }
     }
 }
-

rust/bioscript-cli/src/report_output.rs

@@ -19,6 +19,20 @@ fn app_report_json(input: AppReportJsonInput<'_>) -> serde_json::Value {
             item.get("call_status").and_then(serde_json::Value::as_str) == Some("called")
         })
         .count();
+    let input_debug = input
+        .input_inspection
+        .map(|inspection| {
+            let mut value = input_inspection_json(inspection);
+            if observations_have_imputed_vcf_references(input.observations)
+                && let Some(object) = value.as_object_mut()
+            {
+                object.insert(
+                    "vcf_missing_reference_imputation".to_owned(),
+                    serde_json::Value::Bool(true),
+                );
+            }
+            value
+        });
     serde_json::json!({
         "schema": "bioscript:report:1.0",
         "version": "1.0",
@@ -29,7 +43,7 @@ fn app_report_json(input: AppReportJsonInput<'_>) -> serde_json::Value {
         "input": {
             "file_name": input.input_file.file_name().and_then(|value| value.to_str()).unwrap_or_default(),
             "file_path": input.input_file.display().to_string(),
-            "debug": input.input_inspection.map(input_inspection_json),
+            "debug": input_debug,
         },
         "report_status": if called == input.observations.len() { "complete" } else { "partial" },
         "derived_from": input.observations.iter().filter_map(|item| item.get("variant_key").cloned()).collect::<Vec<_>>(),
@@ -46,6 +60,17 @@ fn app_report_json(input: AppReportJsonInput<'_>) -> serde_json::Value {
     })
 }
 
+fn observations_have_imputed_vcf_references(observations: &[serde_json::Value]) -> bool {
+    observations.iter().any(|observation| {
+        observation
+            .get("evidence_raw")
+            .and_then(serde_json::Value::as_str)
+            .is_some_and(|evidence| {
+                evidence.contains("imputed reference genotype from absent variant-only VCF record")
+            })
+    })
+}
+
 fn report_manifest_metadata(path: &Path) -> Result<serde_json::Value, String> {
     let text = fs::read_to_string(path)
         .map_err(|err| format!("failed to read manifest metadata {}: {err}", path.display()))?;

rust/bioscript-formats/src/genotype.rs

@@ -84,6 +84,10 @@ impl GenotypeStore {
             GenotypeSourceFormat::Zip => Self::from_zip_file(path),
             GenotypeSourceFormat::Vcf => Ok(Self::from_vcf_file(path, options)),
             GenotypeSourceFormat::Cram => Self::from_cram_file(path, options),
+            GenotypeSourceFormat::Bam => Err(RuntimeError::Unsupported(format!(
+                "BAM alignment lookup is not implemented yet for {}",
+                path.display()
+            ))),
         }
     }
 
@@ -361,6 +365,7 @@ impl RsidMapBackend {
             GenotypeSourceFormat::Zip => "zip",
             GenotypeSourceFormat::Vcf => "vcf",
             GenotypeSourceFormat::Cram => "cram",
+            GenotypeSourceFormat::Bam => "bam",
         }
     }
 
@@ -392,6 +397,7 @@ impl DelimitedBackend {
             GenotypeSourceFormat::Zip => "zip",
             GenotypeSourceFormat::Vcf => "vcf",
             GenotypeSourceFormat::Cram => "cram",
+            GenotypeSourceFormat::Bam => "bam",
         }
     }
 
@@ -939,6 +945,18 @@ mod tests {
             detect_source_format(&text, Some(GenotypeSourceFormat::Cram)).unwrap(),
             GenotypeSourceFormat::Cram
         ));
+        let bam = dir.join("sample.bam");
+        fs::write(&bam, b"BAM").unwrap();
+        assert!(matches!(
+            detect_source_format(&bam, None).unwrap(),
+            GenotypeSourceFormat::Bam
+        ));
+        assert!(
+            GenotypeStore::from_file(&bam)
+                .unwrap_err()
+                .to_string()
+                .contains("BAM alignment lookup is not implemented yet")
+        );
         assert!(!looks_like_vcf_lines(&["rsid\tgenotype".to_owned()]));
 
         let backend = DelimitedBackend {

rust/bioscript-formats/src/genotype/cram_backend.rs

@@ -18,6 +18,7 @@ use crate::alignment;
 use super::GenotypeLoadOptions;
 
 mod indel;
+mod observation;
 mod reader;
 mod store;