improvements to reporting

rust/bioscript-cli/src/main.rs

@@ -8,6 +8,7 @@ include!("package.rs");
 include!("report_review.rs");
 include!("report_execution.rs");
 include!("report_observations.rs");
+include!("report_observation_json.rs");
 include!("report_findings.rs");
 include!("report_matching.rs");
 include!("report_output.rs");

rust/bioscript-cli/src/report_findings.rs

@@ -273,22 +273,70 @@ mod report_observations_tests {
         );
     }
 
+    #[test]
+    fn deletion_copy_number_calls_are_normalized_from_insertion_deletion_tokens() {
+        assert_eq!(
+            normalize_app_genotype(
+                "DI",
+                "TTATAA",
+                "<DEL:6>",
+                Some(bioscript_core::VariantKind::Deletion),
+                "22",
+                None
+            ),
+            ("0/1".to_owned(), "het".to_owned())
+        );
+    }
+
+    #[test]
+    fn cram_long_deletion_copy_number_calls_are_displayed_as_insertion_deletion_tokens() {
+        let mut row = BTreeMap::new();
+        row.insert("backend".to_owned(), "cram".to_owned());
+        let manifest = bioscript_schema::VariantManifest {
+            path: std::path::PathBuf::new(),
+            name: "apol1_g2".to_owned(),
+            tags: Vec::new(),
+            spec: bioscript_core::VariantSpec {
+                reference: Some("TTATAA".to_owned()),
+                alternate: Some("<DEL:6>".to_owned()),
+                kind: Some(bioscript_core::VariantKind::Deletion),
+                ..bioscript_core::VariantSpec::default()
+            },
+        };
+
+        assert_eq!(
+            deletion_copy_number_display(&row, &manifest, Some(53), Some(0)).as_deref(),
+            Some("II")
+        );
+        assert_eq!(
+            normalize_app_genotype(
+                "II",
+                "TTATAA",
+                "<DEL:6>",
+                Some(bioscript_core::VariantKind::Deletion),
+                "22",
+                None
+            ),
+            ("0/0".to_owned(), "hom_ref".to_owned())
+        );
+    }
+
     #[test]
     fn single_allele_sex_chromosome_calls_are_treated_as_hemizygous() {
         assert_eq!(
-            normalize_app_genotype("G", "C", "G", "X", None),
+            normalize_app_genotype("G", "C", "G", None, "X", None),
             ("1".to_owned(), "hem_alt".to_owned())
         );
         assert_eq!(
-            normalize_app_genotype("C", "C", "G", "chrX", None),
+            normalize_app_genotype("C", "C", "G", None, "chrX", None),
             ("0".to_owned(), "hem_ref".to_owned())
         );
         assert_eq!(
-            normalize_app_genotype("G", "C", "G", "1", None),
+            normalize_app_genotype("G", "C", "G", None, "1", None),
             ("G".to_owned(), "unknown".to_owned())
         );
         assert_eq!(
-            normalize_app_genotype("GG", "C", "G", "X", None),
+            normalize_app_genotype("GG", "C", "G", None, "X", None),
             ("1/1".to_owned(), "hom_alt".to_owned())
         );
     }
@@ -302,11 +350,11 @@ mod report_observations_tests {
             evidence: vec!["called_y_snps=1200".to_owned()],
         };
         assert_eq!(
-            normalize_app_genotype("GG", "C", "G", "X", Some(&inferred_sex)),
+            normalize_app_genotype("GG", "C", "G", None, "X", Some(&inferred_sex)),
             ("1".to_owned(), "hem_alt".to_owned())
         );
         assert_eq!(
-            normalize_app_genotype("CC", "C", "G", "chrX", Some(&inferred_sex)),
+            normalize_app_genotype("CC", "C", "G", None, "chrX", Some(&inferred_sex)),
             ("0".to_owned(), "hem_ref".to_owned())
         );
     }

rust/bioscript-cli/src/report_html_analysis.rs

@@ -1,6 +1,7 @@
 fn render_analysis_tables(
     out: &mut String,
     analyses: &[serde_json::Value],
+    observations: &[serde_json::Value],
     show_participant_id: bool,
 ) {
     if analyses.is_empty() {
@@ -10,6 +11,7 @@ fn render_analysis_tables(
     for (index, analysis) in analyses.iter().enumerate() {
         let table_id = format!("analysis-table-{index}");
         let title = analysis_title(analysis);
+        let weak_indel_dependency = analysis_depends_on_weak_observation(analysis, observations);
         let row_count = analysis
             .get("rows")
             .and_then(serde_json::Value::as_array)
@@ -37,6 +39,7 @@ fn render_analysis_tables(
             out.push_str("<h4>Results</h4>");
             render_analysis_key_values(out, analysis, &rows[0], &headers);
             render_analysis_notes(out, &notes);
+            render_weak_indel_analysis_note(out, weak_indel_dependency);
             out.push_str("</div></details>");
             continue;
         }
@@ -57,6 +60,7 @@ fn render_analysis_tables(
         }
         render_table_end(out);
         render_analysis_notes(out, &notes);
+        render_weak_indel_analysis_note(out, weak_indel_dependency);
         out.push_str("</div></details>");
     }
 }
@@ -182,7 +186,7 @@ fn render_analysis_value(key: &str, value: &str) -> String {
             html_escape(value)
         );
     }
-    if key.ends_with("_status") || key.ends_with("_outcome") {
+    if is_analysis_badge_key(key) {
         format!(
             "<span class=\"analysis-badge {}\">{}</span>",
             analysis_badge_class(value),
@@ -193,6 +197,10 @@ fn render_analysis_value(key: &str, value: &str) -> String {
     }
 }
 
+fn is_analysis_badge_key(key: &str) -> bool {
+    key.ends_with("_status") || key.ends_with("_outcome")
+}
+
 fn analysis_badge_class(value: &str) -> &'static str {
     match value {
         "normal" | "reference" => "analysis-badge-normal",
@@ -214,6 +222,49 @@ fn render_analysis_notes(out: &mut String, notes: &[String]) {
     out.push_str("</div>");
 }
 
+fn render_weak_indel_analysis_note(out: &mut String, weak_indel_dependency: bool) {
+    if !weak_indel_dependency {
+        return;
+    }
+    out.push_str("<div class=\"analysis-notes\"><h4>Notes</h4><p>* Result depends on a weak indel match from a consumer genotype file.</p></div>");
+}
+
+fn analysis_depends_on_weak_observation(
+    analysis: &serde_json::Value,
+    observations: &[serde_json::Value],
+) -> bool {
+    let weak_paths = observations
+        .iter()
+        .filter(|observation| analysis_observation_is_weak_indel_match(observation))
+        .filter_map(|observation| {
+            observation
+                .get("variant_path")
+                .and_then(serde_json::Value::as_str)
+        })
+        .collect::<Vec<_>>();
+    if weak_paths.is_empty() {
+        return false;
+    }
+    analysis
+        .get("derived_from")
+        .and_then(serde_json::Value::as_array)
+        .into_iter()
+        .flatten()
+        .filter_map(serde_json::Value::as_str)
+        .any(|derived| {
+            weak_paths
+                .iter()
+                .any(|path| path.ends_with(derived) || derived.ends_with(path))
+        })
+}
+
+fn analysis_observation_is_weak_indel_match(observation: &serde_json::Value) -> bool {
+    observation
+        .get("match_quality")
+        .and_then(serde_json::Value::as_str)
+        == Some("weak")
+}
+
 fn render_analysis_logic(out: &mut String, analysis: &serde_json::Value) {
     let Some(logic) = analysis.get("logic") else {
         return;
@@ -249,4 +300,3 @@ fn render_analysis_logic(out: &mut String, analysis: &serde_json::Value) {
     }
     out.push_str("</div>");
 }
-

rust/bioscript-cli/src/report_html_helpers.rs

@@ -32,6 +32,8 @@ fn is_debug_column(header: &str) -> bool {
             | "genotype_quality"
             | "evidence_type"
             | "evidence_raw"
+            | "match_quality"
+            | "match_notes"
             | "assay_id"
             | "assay_version"
             | "variant_key"
@@ -69,6 +71,8 @@ fn table_header_label(header: &str) -> String {
         "evidence_type" => "Evidence Type".to_owned(),
         "source" => "Source".to_owned(),
         "evidence_raw" => "Evidence Raw".to_owned(),
+        "match_quality" => "Match Quality".to_owned(),
+        "match_notes" => "Match Notes".to_owned(),
         "facets" => "Facets".to_owned(),
         "assay_id" => "Assay ID".to_owned(),
         "assay_version" => "Assay Version".to_owned(),

rust/bioscript-cli/src/report_html_observations.rs

@@ -25,6 +25,8 @@ fn render_observation_table(
         "allele_balance",
         "evidence_type",
         "evidence_raw",
+        "match_quality",
+        "match_notes",
         "facets",
         "assay_id",
         "assay_version",
@@ -41,7 +43,12 @@ fn render_observation_table(
     let show_facets = observations
         .iter()
         .any(|observation| !json_field_as_tsv(observation.get("facets")).is_empty());
+    let show_match_quality = observations.iter().any(|observation| {
+        !json_field_as_tsv(observation.get("match_quality")).is_empty()
+            || !json_field_as_tsv(observation.get("match_notes")).is_empty()
+    });
     let show_imputed_reference_note = observations.iter().any(observation_is_imputed_vcf_reference);
+    let show_weak_indel_note = observations.iter().any(observation_is_weak_indel_match);
     let headers = all_headers
         .iter()
         .copied()
@@ -50,6 +57,9 @@ fn render_observation_table(
             show_counts || !matches!(*header, "ref_count" | "alt_count" | "depth" | "allele_balance")
         })
         .filter(|header| show_genotype_quality || *header != "genotype_quality")
+        .filter(|header| {
+            show_match_quality || !matches!(*header, "match_quality" | "match_notes")
+        })
         .filter(|header| show_facets || *header != "facets")
         .collect::<Vec<_>>();
     render_table_start(out, "observations-table", &headers);
@@ -70,6 +80,9 @@ fn render_observation_table(
     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>");
     }
+    if show_weak_indel_note {
+        out.push_str("<p class=\"muted observation-note\">* Indel calls from consumer genotype files are weak matches: the file reports an insertion/deletion token at the marker, but does not provide sequence-resolved evidence for the exact deletion allele.</p>");
+    }
 }
 
 fn observation_filter_group(observation: &serde_json::Value) -> &'static str {
@@ -131,7 +144,9 @@ fn render_observation_cell(out: &mut String, observation: &serde_json::Value, he
     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) {
+        if (value == "reference" && observation_is_imputed_vcf_reference(observation))
+            || (value == "variant" && observation_is_weak_indel_match(observation))
+        {
             value.push('*');
         }
         let _ = write!(out, "<td class=\"{}\">{}</td>", cell_class, html_escape(&value));
@@ -205,6 +220,13 @@ fn observation_is_imputed_vcf_reference(observation: &serde_json::Value) -> bool
         })
 }
 
+fn observation_is_weak_indel_match(observation: &serde_json::Value) -> bool {
+    observation
+        .get("match_quality")
+        .and_then(serde_json::Value::as_str)
+        == Some("weak")
+}
+
 fn observation_ref_alt(observation: &serde_json::Value) -> String {
     let ref_allele = observation
         .get("ref")