Madhava/fixes vc

rust/bioscript-formats/src/genotype.rs

@@ -10,6 +10,7 @@ use zip::ZipArchive;
 use bioscript_core::{RuntimeError, VariantObservation, VariantSpec};
 
 mod backends;
+mod cache;
 mod common;
 mod cram_backend;
 mod delimited;
@@ -19,6 +20,7 @@ mod types;
 mod vcf;
 mod vcf_tokens;
 
+pub(crate) use cache::{match_cached_observation, required_cache_miss};
 pub(crate) use common::{describe_query, normalize_genotype, variant_sort_key};
 pub use cram_backend::{
     observe_cram_deletion_with_reader, observe_cram_indel_with_reader, observe_cram_snp_with_reader,
@@ -27,14 +29,17 @@ pub(crate) use delimited::{
     COMMENT_PREFIXES, DelimitedColumnIndexes, Delimiter, detect_delimiter, parse_streaming_row,
 };
 use delimited::{RowParser, scan_delimited_variants};
-use io::{detect_source_format, is_bgzf_path, read_lines_from_reader, select_zip_entry};
+use io::{
+    detect_source_format, is_bgzf_path, looks_like_vcf_lines, read_lines_from_reader,
+    select_zip_entry,
+};
 pub use types::{
     BackendCapabilities, GenotypeLoadOptions, GenotypeSourceFormat, GenotypeStore, QueryKind,
 };
 use types::{CramBackend, DelimitedBackend, QueryBackend, RsidMapBackend, VcfBackend};
 pub use vcf::{
-    choose_variant_locus_for_assembly, imputed_reference_observation, observe_vcf_snp_with_reader,
-    observe_vcf_variant_with_reader,
+    choose_variant_locus_for_assembly, detect_vcf_assembly, imputed_reference_observation,
+    observe_vcf_snp_with_reader, observe_vcf_variant_with_reader,
 };
 use vcf::{lookup_indexed_vcf_variants, scan_vcf_variants};
 pub(crate) use vcf_tokens::genotype_from_vcf_gt;
@@ -119,12 +124,16 @@ impl GenotypeStore {
     }
 
     pub fn from_bytes(name: &str, bytes: &[u8]) -> Result<Self, RuntimeError> {
+        // The report pipeline hands us a fixed virtual path (`/input/genotypes`)
+        // with no extension, so we cannot rely on `name` alone for format
+        // detection the way `from_file_with_options` can. Sniff the leading
+        // bytes so a zip/VCF payload is recognised regardless of the name.
         let lower = name.to_ascii_lowercase();
-        if lower.ends_with(".zip") {
+        if lower.ends_with(".zip") || bytes_look_like_zip(bytes) {
             return Self::from_zip_bytes(name, bytes);
         }
         let reader = BufReader::new(Cursor::new(bytes));
-        if lower.ends_with(".vcf") {
+        if lower.ends_with(".vcf") || bytes_look_like_vcf(bytes) {
             return Self::from_vcf_reader(reader, name);
         }
         Self::from_delimited_reader(GenotypeSourceFormat::Text, reader, name)
@@ -425,68 +434,17 @@ impl GenotypeStore {
     }
 }
 
-/// Match a `VariantSpec` against a pre-resolved observation list. Tries rsid
-/// equality first (most common case for `PGx` panels), then falls back to a
-/// chrom+pos+ref+alt match against either `GRCh37` or `GRCh38` loci so cached
-/// observations from a CRAM lookup (which may have been done on one assembly)
-/// can satisfy a script that supplies the spec on the other.
-fn match_cached_observation<'a>(
-    observations: &'a [VariantObservation],
-    spec: &VariantSpec,
-) -> Option<&'a VariantObservation> {
-    if let Some(matched) = observations.iter().find(|obs| {
-        obs.matched_rsid
-            .as_deref()
-            .is_some_and(|rsid| spec.rsids.iter().any(|target| target == rsid))
-    }) {
-        return Some(matched);
-    }
-    let assembly_loci = [spec.grch37.as_ref(), spec.grch38.as_ref()]
-        .into_iter()
-        .flatten()
-        .collect::<Vec<_>>();
-    let target_ref = spec.reference.as_deref();
-    let target_alt = spec.alternate.as_deref();
-    observations.iter().find(|obs| {
-        let evidence_match = assembly_loci.iter().any(|loci| {
-            obs.evidence
-                .iter()
-                .any(|line| line.contains(&loci.chrom) && line.contains(&loci.start.to_string()))
-        });
-        if !evidence_match {
-            return false;
-        }
-        match (target_ref, target_alt) {
-            (Some(r), Some(a)) => obs
-                .evidence
-                .iter()
-                .any(|line| line.contains(r) && line.contains(a)),
-            _ => true,
-        }
-    })
+fn bytes_look_like_zip(bytes: &[u8]) -> bool {
+    bytes.starts_with(b"PK\x03\x04")
+        || bytes.starts_with(b"PK\x05\x06")
+        || bytes.starts_with(b"PK\x07\x08")
 }
 
-fn required_cache_miss(spec: &VariantSpec) -> RuntimeError {
-    let rsids = if spec.rsids.is_empty() {
-        "<none>".to_owned()
-    } else {
-        spec.rsids.join("|")
-    };
-    let loci = [
-        spec.grch37
-            .as_ref()
-            .map(|locus| format!("grch37:{}:{}-{}", locus.chrom, locus.start, locus.end)),
-        spec.grch38
-            .as_ref()
-            .map(|locus| format!("grch38:{}:{}-{}", locus.chrom, locus.start, locus.end)),
-    ]
-    .into_iter()
-    .flatten()
-    .collect::<Vec<_>>()
-    .join(",");
-    RuntimeError::InvalidArguments(format!(
-        "required preloaded genotype observation missing for rsids={rsids} loci={loci}"
-    ))
+fn bytes_look_like_vcf(bytes: &[u8]) -> bool {
+    let prefix = &bytes[..bytes.len().min(8192)];
+    let text = String::from_utf8_lossy(prefix);
+    let lines: Vec<String> = text.lines().map(str::to_owned).collect();
+    looks_like_vcf_lines(&lines)
 }
 
 #[cfg(test)]

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

@@ -0,0 +1,65 @@
+use bioscript_core::{RuntimeError, VariantObservation, VariantSpec};
+
+/// Match a `VariantSpec` against a pre-resolved observation list. Tries rsid
+/// equality first (most common case for `PGx` panels), then falls back to a
+/// chrom+pos+ref+alt match against either `GRCh37` or `GRCh38` loci so cached
+/// observations from a CRAM lookup (which may have been done on one assembly)
+/// can satisfy a script that supplies the spec on the other.
+pub(crate) fn match_cached_observation<'a>(
+    observations: &'a [VariantObservation],
+    spec: &VariantSpec,
+) -> Option<&'a VariantObservation> {
+    if let Some(matched) = observations.iter().find(|obs| {
+        obs.matched_rsid
+            .as_deref()
+            .is_some_and(|rsid| spec.rsids.iter().any(|target| target == rsid))
+    }) {
+        return Some(matched);
+    }
+    let assembly_loci = [spec.grch37.as_ref(), spec.grch38.as_ref()]
+        .into_iter()
+        .flatten()
+        .collect::<Vec<_>>();
+    let target_ref = spec.reference.as_deref();
+    let target_alt = spec.alternate.as_deref();
+    observations.iter().find(|obs| {
+        let evidence_match = assembly_loci.iter().any(|loci| {
+            obs.evidence
+                .iter()
+                .any(|line| line.contains(&loci.chrom) && line.contains(&loci.start.to_string()))
+        });
+        if !evidence_match {
+            return false;
+        }
+        match (target_ref, target_alt) {
+            (Some(r), Some(a)) => obs
+                .evidence
+                .iter()
+                .any(|line| line.contains(r) && line.contains(a)),
+            _ => true,
+        }
+    })
+}
+
+pub(crate) fn required_cache_miss(spec: &VariantSpec) -> RuntimeError {
+    let rsids = if spec.rsids.is_empty() {
+        "<none>".to_owned()
+    } else {
+        spec.rsids.join("|")
+    };
+    let loci = [
+        spec.grch37
+            .as_ref()
+            .map(|locus| format!("grch37:{}:{}-{}", locus.chrom, locus.start, locus.end)),
+        spec.grch38
+            .as_ref()
+            .map(|locus| format!("grch38:{}:{}-{}", locus.chrom, locus.start, locus.end)),
+    ]
+    .into_iter()
+    .flatten()
+    .collect::<Vec<_>>()
+    .join(",");
+    RuntimeError::InvalidArguments(format!(
+        "required preloaded genotype observation missing for rsids={rsids} loci={loci}"
+    ))
+}

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

@@ -429,6 +429,6 @@ pub(crate) fn extract_vcf_sample_genotype(
     genotype_from_vcf_gt(sample_gt, reference, &alternate_refs)
 }
 
-pub(crate) fn detect_vcf_assembly(path: &Path, probe_lines: &[String]) -> Option<Assembly> {
+pub fn detect_vcf_assembly(path: &Path, probe_lines: &[String]) -> Option<Assembly> {
     detect_assembly(&path.to_string_lossy().to_ascii_lowercase(), probe_lines)
 }

rust/bioscript-formats/src/lib.rs

@@ -15,7 +15,7 @@ mod prepare;
 
 pub use genotype::{
     BackendCapabilities, GenotypeLoadOptions, GenotypeSourceFormat, GenotypeStore, QueryKind,
-    choose_variant_locus_for_assembly, imputed_reference_observation,
+    choose_variant_locus_for_assembly, detect_vcf_assembly, imputed_reference_observation,
     observe_cram_deletion_with_reader, observe_cram_indel_with_reader,
     observe_cram_snp_with_reader, observe_vcf_snp_with_reader, observe_vcf_variant_with_reader,
 };

rust/bioscript-reporting/src/html/helpers.rs

@@ -3,7 +3,7 @@ pub(super) fn render_table_start(out: &mut String, table_id: &str, headers: &[&s
     let escaped_id = html_escape(table_id);
     let _ = write!(
         out,
-        "<div class=\"table-tools\"><input type=\"search\" placeholder=\"Filter table\" data-filter-for=\"{escaped_id}\" oninput=\"applyTableFilters('{escaped_id}')\"></div><div class=\"table-wrap\"><table id=\"{escaped_id}\"><thead><tr>"
+        "<div class=\"table-tools\"><input type=\"search\" placeholder=\"Filter table\" data-filter-for=\"{escaped_id}\" oninput=\"scheduleTableFilter('{escaped_id}')\"></div><div class=\"table-wrap\"><table id=\"{escaped_id}\"><thead><tr>"
     );
     for (index, header) in headers.iter().enumerate() {
         let _ = write!(

rust/bioscript-reporting/src/manifest.rs

@@ -1,4 +1,5 @@
 use std::{
+    collections::BTreeSet,
     fs,
     path::{Path, PathBuf},
 };
@@ -284,6 +285,7 @@ pub fn collect_variant_manifest_tasks(
                     }
                 }
             }
+            dedupe_variant_manifest_tasks(&mut tasks);
             Ok(tasks)
         }
         ReportManifestKind::Assay => {
@@ -305,11 +307,17 @@ pub fn collect_variant_manifest_tasks(
                     }
                 }
             }
+            dedupe_variant_manifest_tasks(&mut tasks);
             Ok(tasks)
         }
     }
 }
 
+fn dedupe_variant_manifest_tasks(tasks: &mut Vec<VariantManifestTask>) {
+    let mut seen = BTreeSet::new();
+    tasks.retain(|task| seen.insert(task.manifest_path.clone()));
+}
+
 fn load_variant_task(
     workspace: &impl ManifestWorkspace,
     path: &str,
@@ -804,6 +812,62 @@ analyses:
         assert_eq!(variant_tasks[0].manifest.name, "rs1");
     }
 
+    #[test]
+    fn collect_variant_manifest_tasks_dedupes_catalogue_reached_through_panel_and_assay() {
+        let workspace = MapWorkspace {
+            files: BTreeMap::from([
+                (
+                    "panel.yaml".to_owned(),
+                    r#"
+schema: bioscript:panel:1.0
+version: "1.0"
+name: panel
+members:
+  - kind: variant-catalogue
+    path: catalogue.yaml
+  - kind: assay
+    path: assay.yaml
+"#
+                    .to_owned(),
+                ),
+                (
+                    "assay.yaml".to_owned(),
+                    r#"
+schema: bioscript:assay:1.0
+version: "1.0"
+name: assay
+members:
+  - kind: variant-catalogue
+    path: catalogue.yaml
+"#
+                    .to_owned(),
+                ),
+                (
+                    "catalogue.yaml".to_owned(),
+                    r#"
+schema: bioscript:variant-catalogue:1.0
+version: "1.0"
+name: catalogue
+variants:
+  source: variants.tsv
+"#
+                    .to_owned(),
+                ),
+                (
+                    "variants.tsv".to_owned(),
+                    r#"id	name	rsid	gene	ref	alt	kind	grch38_chrom	grch38_pos
+rs1	rs1	rs1	GENE	A	G	snp	1	123
+"#
+                    .to_owned(),
+                ),
+            ]),
+        };
+
+        let tasks = collect_variant_manifest_tasks(&workspace, "panel.yaml", &[]).unwrap();
+        assert_eq!(tasks.len(), 1);
+        assert_eq!(tasks[0].manifest_path, "catalogue.yaml#rs1");
+    }
+
     #[test]
     fn manifest_context_and_findings_follow_includes_members_and_inherited_bindings() {
         let workspace = MapWorkspace {