Add distribution endpoints and batch retrieval

bedhost/cli.py

@@ -3,6 +3,8 @@
 from . import PKG_NAME
 from ._version import __version__
 
+CFG_ENV_VARS = ["BEDBASE_CONFIG"]
+
 
 def build_parser():
     """

bedhost/const.py

@@ -15,6 +15,8 @@
 USAGE_RECORD_DAYS = 30
 
 
+MAX_BATCH_IDS = 1000
+
 MAX_FILE_SIZE = 1024 * 1024 * 2
 MAX_REGION_NUMBER = 5000000
 MIN_REGION_WIDTH = 10

bedhost/data_models.py

@@ -84,5 +84,13 @@ class CreateBEDsetRequest(BaseModel):
     registry_path: str
 
 
+class BatchBedRequest(BaseModel):
+    ids: list[str]
+
+
+class CollectionStatsRequest(BaseModel):
+    ids: list[str]
+
+
 class ChromLengthUploadModel(BaseModel):
     bed_file: Dict[str, int]

bedhost/routers/base_api.py

@@ -6,7 +6,12 @@
 
 from platform import python_version
 
-from bbconf import __version__ as bbconf_version
+try:
+    from bbconf import __version__ as bbconf_version
+except ImportError:
+    from importlib.metadata import version
+
+    bbconf_version = version("bbconf")
 from bbconf.models.base_models import StatsReturn, FileStats, UsageStats
 from fastapi import APIRouter, Request
 from geniml import __version__ as geniml_version

bedhost/routers/bed_api.py

@@ -36,12 +36,20 @@
 from gtars.models import RegionSet
 
 from .. import _LOGGER
-from ..const import EXAMPLE_BED, MAX_FILE_SIZE, MAX_REGION_NUMBER, MIN_REGION_WIDTH
+from ..const import (
+    EXAMPLE_BED,
+    MAX_BATCH_IDS,
+    MAX_FILE_SIZE,
+    MAX_REGION_NUMBER,
+    MIN_REGION_WIDTH,
+)
 from ..data_models import (
     CROM_NUMBERS,
     BaseListResponse,
+    BatchBedRequest,
     BedDigest,
     ChromLengthUploadModel,
+    CollectionStatsRequest,
 )
 from ..main import bbagent, usage_data, ref_validator
 from ..helpers import count_requests, test_query_parameter
@@ -159,15 +167,51 @@ async def get_bed_files(
 )
 async def get_bed_stats(
     bed_id: str = BedDigest,
+    distributions: bool = Query(
+        True,
+        description="Include distribution arrays in the response. Set to false to exclude the large distributions blob.",
+    ),
 ):
     try:
-        return bbagent.bed.get_stats(bed_id)
+        return bbagent.bed.get_stats(bed_id, distributions=distributions)
     except BEDFileNotFoundError as _:
         raise HTTPException(
             status_code=404,
         )
 
 
+@router.post(
+    "/batch",
+    summary="Get metadata for multiple BED records",
+)
+async def get_bed_batch(
+    request: BatchBedRequest,
+):
+    """Retrieve metadata for multiple BED files in a single request."""
+    if len(request.ids) > MAX_BATCH_IDS:
+        raise HTTPException(
+            status_code=400,
+            detail=f"Too many IDs. Maximum is {MAX_BATCH_IDS}.",
+        )
+    return bbagent.bed.get_batch(request.ids, full=True, distributions=False)
+
+
+@router.post(
+    "/collection/stats",
+    summary="Compute ad-hoc collection statistics from a list of BED IDs",
+)
+async def get_collection_stats(
+    request: CollectionStatsRequest,
+):
+    """Aggregate distribution statistics across a set of BED files on the fly."""
+    if len(request.ids) > MAX_BATCH_IDS:
+        raise HTTPException(
+            status_code=400,
+            detail=f"Too many IDs. Maximum is {MAX_BATCH_IDS}.",
+        )
+    return bbagent.bed.aggregate_collection(request.ids)
+
+
 @router.get(
     "/{bed_id}/metadata/classification",
     summary="Get classification of single BED file",

bedhost/routers/bedset_api.py

@@ -3,6 +3,7 @@
 from bbconf.exceptions import BedSetNotFoundError, BedSetTrackHubLimitError
 from bbconf.models.bedset_models import (
     BedSetBedFiles,
+    BedSetDistributions,
     BedSetListResult,
     BedSetMetadata,
     BedSetPlots,
@@ -109,10 +110,10 @@ async def get_bedset_metadata(
 @router.get(
     "/{bedset_id}/metadata/stats",
     response_model=BedSetStats,
-    summary="Get stats for a single BEDSET record",
-    description=f"Example\n bed_id: {EXAMPLE_BEDSET}",
+    summary="Get scalar stats (mean/sd) for a single BEDSET record",
+    description=f"Example\n bedset_id: {EXAMPLE_BEDSET}",
 )
-async def get_bedset_metadata(
+async def get_bedset_stats(
     bedset_id: str,
 ):
     try:
@@ -121,6 +122,27 @@ async def get_bedset_metadata(
         raise HTTPException(status_code=404, detail="No records found")
 
 
+@router.get(
+    "/{bedset_id}/metadata/distributions",
+    response_model=BedSetDistributions,
+    summary="Get distribution statistics for a single BEDSET record",
+    description=f"Example\n bedset_id: {EXAMPLE_BEDSET}",
+)
+async def get_bedset_distributions(
+    bedset_id: str,
+):
+    """Get aggregated distribution statistics from the JSONB column.
+
+    Returns full distribution data (histograms, KDEs, partitions, etc.)
+    computed across all member BED files. Falls back to wrapping old
+    scalar stats if distribution data is not yet available.
+    """
+    try:
+        return bbagent.bedset.get_distributions(bedset_id)
+    except BedSetNotFoundError as _:
+        raise HTTPException(status_code=404, detail="No records found")
+
+
 @router.get(
     "/{bedset_id}/bedfiles",
     response_model=BedSetBedFiles,

ui/package.json

@@ -9,7 +9,9 @@
     "lint": "eslint . --ext ts,tsx --report-unused-disable-directives --max-warnings 0",
     "preview": "vite preview",
     "generate-types": "npx openapi-typescript https://api-dev.bedbase.org/openapi.json -o bedbase-types.d.ts",
-    "generate-types-local": "npx openapi-typescript http://localhost:8000/openapi.json -o bedbase-types.d.ts"
+    "generate-types-local": "npx openapi-typescript http://localhost:8000/openapi.json -o bedbase-types.d.ts",
+    "deploy": "npm run build && wrangler deploy",
+    "deploy:preview": "npm run build && wrangler dev"
   },
   "dependencies": {
     "@databio/gtars": "^0.5.3",

ui/wrangler.jsonc

@@ -0,0 +1,8 @@
+{
+  "name": "bedhost-ui",
+  "compatibility_date": "2026-02-20",
+  "assets": {
+    "directory": "./dist",
+    "not_found_handling": "single-page-application"
+  }
+}