Add modular analysis backend config and distribution schema

bbconf/config_parser/models.py

@@ -1,5 +1,6 @@
 import logging
 from pathlib import Path
+from typing import Literal
 
 from pydantic import BaseModel, ConfigDict, computed_field, field_validator
 from yacman import load_yaml
@@ -126,6 +127,17 @@ class ConfigPepHubClient(BaseModel):
     tag: str | None = DEFAULT_PEPHUB_TAG
 
 
+class ConfigAnalysis(BaseModel):
+    """Analysis backend configuration.
+
+    Controls which statistics engine is used for BED file analysis.
+    """
+
+    backend: Literal["r", "gtars"] = "r"
+
+    model_config = ConfigDict(extra="forbid")
+
+
 class ConfigFile(BaseModel):
     database: ConfigDB
     qdrant: ConfigQdrant = None
@@ -134,6 +146,7 @@ class ConfigFile(BaseModel):
     access_methods: AccessMethods = None
     s3: ConfigS3 = None
     phc: ConfigPepHubClient = None
+    analysis: ConfigAnalysis = ConfigAnalysis()
 
     model_config = ConfigDict(extra="allow")
 

bbconf/db_utils.py

@@ -255,6 +255,12 @@ class BedStats(Base):
     promotercore_percentage: Mapped[Optional[float]]
     tssdist: Mapped[Optional[float]]
 
+    distributions: Mapped[Optional[dict]] = mapped_column(
+        JSON,
+        nullable=True,
+        comment="Full distribution arrays from gtars genomicdist (JSONB)",
+    )
+
     bed: Mapped["Bed"] = relationship("Bed", back_populates="stats")
 
 
@@ -337,6 +343,11 @@ class BedSets(Base):
     bedset_standard_deviation: Mapped[Optional[dict]] = mapped_column(
         JSON, comment="Median values of the bedset"
     )
+    bedset_stats: Mapped[Optional[dict]] = mapped_column(
+        JSON,
+        nullable=True,
+        comment="Pre-aggregated distribution statistics from gtars (JSONB)",
+    )
 
     bedfiles: Mapped[list["BedFileBedSetRelation"]] = relationship(
         "BedFileBedSetRelation", back_populates="bedset", cascade="all, delete-orphan"

bbconf/models/bed_models.py

@@ -74,6 +74,8 @@ class BedStatsModel(BaseModel):
     promoterprox_frequency: float | None = None
     promoterprox_percentage: float | None = None
 
+    distributions: dict | None = None
+
     model_config = ConfigDict(extra="ignore", populate_by_name=True)
 
 

bbconf/models/bedset_models.py

@@ -1,4 +1,5 @@
 import datetime
+from typing import Optional
 
 from pydantic import BaseModel, ConfigDict, model_validator
 
@@ -7,10 +8,34 @@
 
 
 class BedSetStats(BaseModel):
+    """Bedset statistics: mean/sd of scalar columns.
+
+    Populated from bedset_means and bedset_standard_deviation database columns.
+    """
+
     mean: BedStatsModel = None
     sd: BedStatsModel = None
 
 
+class BedSetDistributions(BaseModel):
+    """Collection-level aggregated distribution statistics for a bedset.
+
+    Stored in the bedset_stats JSONB database column. Populated when
+    member bed files have been processed with the gtars analysis backend.
+    """
+
+    n_files: int = 0
+    composition: Optional[dict] = None
+    scalar_summaries: Optional[dict] = None
+    tss_histogram: Optional[dict] = None
+    widths_histogram: Optional[dict] = None
+    neighbor_distances: Optional[dict] = None
+    gc_content: Optional[dict] = None
+    region_distribution: Optional[dict] = None
+    partitions: Optional[dict] = None
+    chromosome_summaries: Optional[dict] = None
+
+
 class BedSetPlots(BaseModel):
     region_commonality: FileModel = None
 
@@ -24,6 +49,7 @@ class BedSetMetadata(BaseModel):
     submission_date: datetime.datetime = None
     last_update_date: datetime.datetime = None
     statistics: BedSetStats | None = None
+    distributions: BedSetDistributions | None = None
     plots: BedSetPlots | None = None
     description: str = None
     summary: str = None

bbconf/modules/bedsets.py

@@ -427,7 +427,11 @@ def _calculate_statistics(self, bed_ids: list[str]) -> BedSetStats:
         """
 
         _LOGGER.info("Calculating bedset statistics")
-        numeric_columns = BedStatsModel.model_fields
+        numeric_columns = [
+            name
+            for name, field in BedStatsModel.model_fields.items()
+            if field.annotation in (float, float | None)
+        ]
 
         bedset_sd = {}
         bedset_mean = {}