Release v1.0.2 (#12)

CHANGELOG.md

@@ -1,5 +1,18 @@
 # Changelog
 
+## [1.0.2] — 2026-05-25
+
+### Fixed
+
+- **DELLY writer**: `FORMAT` field order corrected from `…FT:RC:RCL:RCR…` to `…FT:RCL:RC:RCR…` to match real DELLY 1.x output; sample column values reordered accordingly so `RC` stays the SV-region count and `RCL`/`RCR` the left/right control regions.
+- **Manta writer**: DUP records now emit `<DUP:TANDEM>` as ALT (matching the `##ALT` declaration in the Manta header) instead of the undeclared `<DUP>`.
+- **DELLY writer**: BND records now use the standard VCF 4.2 breakend notation (e.g. `]chr3:151368767]C`) instead of the symbolic `<BND>` allele, matching real DELLY output.
+- **Sampler**: SV anchor `REF` base is now drawn uniformly from A/C/G/T per record instead of being hardcoded to `N`. This makes BND ALT strings look like `G]chr1:42]` instead of `N]chr1:42]`.
+
+### Added
+
+- README "Scope & limitations" section documenting which INFO/FORMAT fields are declared in the bundled headers but not (yet) populated, split by caller (Manta vs DELLY).
+
 ## [1.0.1] — 2026-04-28
 
 ### Fixed

README.md

@@ -37,6 +37,17 @@ pip install -e ".[dev,test]"
 
 For ready-to-run command lines (single-sample `gen`, paired somatic `gen-pair`, validation, banks, and dev checks), see [`docs/ready-to-use.md`](docs/ready-to-use.md).
 
+## Scope & limitations
+
+svForge ships the full header of each target caller verbatim, but for now it does not populate every INFO/FORMAT field a real run would emit.
+The fields below are declared in the header for parser compatibility but left empty in records — support may land in a later release.
+
+**Manta** — not yet populated: `CIPOS`, `CIEND` (alignment-derived breakpoint CIs), `HOMSEQ` (microhomology sequence; `HOMLEN` is populated), `BND_DEPTH`, `MATE_BND_DEPTH` (per-breakend coverage).
+
+**DELLY** — not yet populated: `RDRATIO` (tumor/normal read-depth ratio), `INSLEN`, `SRMAPQ`, `CONSENSUS`, `CONSBP` (split-read consensus); also, records are currently always tagged `PRECISE` rather than alternating with `IMPRECISE`.
+
+In the meantime, if your downstream pipeline hard-requires any of these fields, you can supply your own `--header-template` that drops the unused declarations, or post-process the VCF to fill them with sentinel values.
+
 ## Typical use cases
 
 - Validate downstream filters (for example, `SVFORGE_SOURCE=gnomad` records should disappear after your gnomAD filtering step).

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "svforge"
-version = "1.0.1"
+version = "1.0.2"
 description = "Synthetic VCF generator for structural variants (Manta, DELLY) with controlled variability and realistic artefact injection"
 readme = "README.md"
 requires-python = ">=3.10"

src/svforge/core/sampler.py

@@ -38,6 +38,19 @@
 
 _CHROM_SAMPLE_MARGIN = 10_000
 _BLACKLIST_SVTYPES = ("DEL", "DUP", "INV")
+_REF_BASES = ("A", "C", "G", "T")
+
+
+def _random_ref_base(rng: random.Random) -> str:
+    """
+    Return a plausible REF base for an SV anchor (random A/C/G/T)
+
+    Real callers always emit a base extracted from the reference fasta. svforge
+    has no fasta access, so it draws uniformly from the four canonical bases.
+    This keeps the BND ALT (which embeds REF) looking like e.g. ``G]chr1:42]``
+    instead of ``N]chr1:42]``
+    """
+    return rng.choice(_REF_BASES)
 
 _T = TypeVar("_T")
 
@@ -294,6 +307,7 @@ def _gnomad_entry_to_sv(entry: GnomadEntry, cfg: SamplerConfig, rng: random.Rand
             strands="+-",
             vaf=vaf,
             genotype=genotype,
+            ref_base=_random_ref_base(rng),
             origin=cfg.origin,
             source="gnomad",
             info_extra={"SVFORGE_SOURCE_ID": entry.source_id},
@@ -310,6 +324,7 @@ def _gnomad_entry_to_sv(entry: GnomadEntry, cfg: SamplerConfig, rng: random.Rand
         strands=_default_strands(entry.svtype),
         vaf=vaf,
         genotype=genotype,
+        ref_base=_random_ref_base(rng),
         origin=cfg.origin,
         source="gnomad",
         info_extra={"SVFORGE_SOURCE_ID": entry.source_id},
@@ -338,6 +353,7 @@ def _blacklist_entry_to_sv(entry: BlacklistEntry, cfg: SamplerConfig, rng: rando
         strands=_default_strands(svtype),
         vaf=vaf,
         genotype=genotype,
+        ref_base=_random_ref_base(rng),
         origin=cfg.origin,
         source="blacklist",
         info_extra={"SVFORGE_SOURCE_ID": entry.source_id},
@@ -395,6 +411,7 @@ def _materialize(
             homlen=homlen,
             vaf=vaf,
             genotype=genotype,
+            ref_base=_random_ref_base(rng),
             origin=cfg.origin,
             source="bank",
         )
@@ -418,6 +435,7 @@ def _materialize(
         homlen=homlen,
         vaf=vaf,
         genotype=genotype,
+        ref_base=_random_ref_base(rng),
         origin=cfg.origin,
         source="bank",
     )

src/svforge/writers/delly.py

@@ -68,17 +68,17 @@ def _paired_sample_columns(sv: SV) -> tuple[str, str, str]:
         if sv.genotype == "1/1"
         else ("0,-5,-50" if sv.genotype == "0/0" else "-10,-1,-10")
     )
-    fmt = "GT:GL:GQ:FT:RC:RCL:RCR:RDCN:DR:DV:RR:RV"
+    fmt = "GT:GL:GQ:FT:RCL:RC:RCR:RDCN:DR:DV:RR:RV"
 
     tumor_col = (
-        f"{sv.genotype}:{gl_alt}:{gq}:PASS:40:20:20:2:"
+        f"{sv.genotype}:{gl_alt}:{gq}:PASS:20:40:20:2:"
         f"{ref}:{alt}:{junction_ref}:{junction_alt}"
     )
 
     if sv.origin == "germline":
         normal_col = tumor_col
     else:
-        normal_col = f"0/0:0,-5,-50:{gq}:PASS:40:20:20:2:30:0:15:0"
+        normal_col = f"0/0:0,-5,-50:{gq}:PASS:20:40:20:2:30:0:15:0"
 
     return fmt, tumor_col, normal_col
 
@@ -117,8 +117,8 @@ def _sample_column(sv: SV) -> tuple[str, str]:
         if sv.genotype == "1/1"
         else ("0,-5,-50" if sv.genotype == "0/0" else "-10,-1,-10")
     )
-    fmt = "GT:GL:GQ:FT:RC:RCL:RCR:RDCN:DR:DV:RR:RV"
-    sample = f"{sv.genotype}:{gl}:{gq}:PASS:40:20:20:2:{ref}:{alt}:{junction_ref}:{junction_alt}"
+    fmt = "GT:GL:GQ:FT:RCL:RC:RCR:RDCN:DR:DV:RR:RV"
+    sample = f"{sv.genotype}:{gl}:{gq}:PASS:20:40:20:2:{ref}:{alt}:{junction_ref}:{junction_alt}"
     return fmt, sample
 
 
@@ -155,8 +155,35 @@ def _base_info(sv: SV) -> list[str]:
     return info
 
 
+def _bnd_alt(ref_base: str, mate_chrom: str, mate_pos: int, strands: str) -> str:
+    """
+    Return the VCF 4.2 breakend ALT string used by real DELLY for BND records
+
+    Real DELLY emits the standard breakend notation (e.g. ``]chr3:151368767]C``)
+    rather than the ``<BND>`` symbolic allele. The ``<BND>`` declaration in the
+    bundled header is kept verbatim from the reference template for header
+    fidelity but is never actually consumed by Delly's record format
+    """
+    loc = f"{mate_chrom}:{mate_pos}"
+    s1, s2 = strands[0], strands[1]
+    if s1 == "+" and s2 == "+":
+        return f"{ref_base}[{loc}["
+    if s1 == "+" and s2 == "-":
+        return f"{ref_base}]{loc}]"
+    if s1 == "-" and s2 == "+":
+        return f"[{loc}[{ref_base}"
+    return f"]{loc}]{ref_base}"
+
+
+def _delly_alt(sv: SV) -> str:
+    if sv.svtype == "BND":
+        assert sv.mate_chrom is not None and sv.mate_pos is not None
+        return _bnd_alt(sv.ref_base, sv.mate_chrom, sv.mate_pos, sv.strands)
+    return f"<{sv.svtype}>"
+
+
 def _delly_record(sv: SV) -> VCFRecord:
-    alt = f"<{sv.svtype}>"
+    alt = _delly_alt(sv)
     info = ";".join(_base_info(sv))
     fmt, sample = _sample_column(sv)
     line = "\t".join(
@@ -177,7 +204,7 @@ def _delly_record(sv: SV) -> VCFRecord:
 
 
 def _delly_record_paired(sv: SV) -> VCFRecord:
-    alt = f"<{sv.svtype}>"
+    alt = _delly_alt(sv)
     info = ";".join(_base_info(sv))
     fmt, tumor_col, normal_col = _paired_sample_columns(sv)
     line = "\t".join(

src/svforge/writers/manta.py

@@ -124,8 +124,20 @@ def _base_info(sv: SV) -> list[str]:
     return info
 
 
+def _symbolic_alt(svtype: str) -> str:
+    """
+    Return the symbolic ALT used by Manta for ``svtype``
+
+    Manta declares ``<DUP:TANDEM>`` (not ``<DUP>``) in its header ALT meta and
+    every DUP record in the wild carries that exact symbolic value
+    """
+    if svtype == "DUP":
+        return "<DUP:TANDEM>"
+    return f"<{svtype}>"
+
+
 def _symbolic_record(sv: SV) -> VCFRecord:
-    alt = f"<{sv.svtype}>"
+    alt = _symbolic_alt(sv.svtype)
     info = ";".join(_base_info(sv))
     fmt, sample = _sample_column(sv)
     line = "\t".join(
@@ -146,7 +158,7 @@ def _symbolic_record(sv: SV) -> VCFRecord:
 
 
 def _symbolic_record_paired(sv: SV) -> VCFRecord:
-    alt = f"<{sv.svtype}>"
+    alt = _symbolic_alt(sv.svtype)
     info = ";".join(_base_info(sv))
     fmt, normal_col, tumor_col = _paired_sample_columns(sv)
     line = "\t".join(