Release 0.12.2

bedhost/_version.py

@@ -1 +1 @@
-__version__ = "0.12.1"
+__version__ = "0.12.2"

bedhost/data_models.py

@@ -1,4 +1,5 @@
 from enum import Enum
+from typing import Dict
 
 from fastapi import Path
 from pydantic import BaseModel
@@ -81,3 +82,7 @@ class BaseListResponse(BaseModel):
 
 class CreateBEDsetRequest(BaseModel):
     registry_path: str
+
+
+class ChromLengthUploadModel(BaseModel):
+    bed_file: Dict[str, int]

bedhost/main.py

@@ -10,6 +10,8 @@
     MissingObjectError,
     MissingThumbnailError,
 )
+from bedboss.refgenome_validator.main import ReferenceValidator
+
 from fastapi import FastAPI, Request
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import HTMLResponse, RedirectResponse
@@ -142,6 +144,11 @@ def main():
 
         global bbagent
         bbagent = configure(bbconf_file_path)
+
+        _LOGGER.info("Initializing reference genome validator...")
+        global ref_validator
+        ref_validator = ReferenceValidator()
+
         attach_routers(app)
         uvicorn.run(
             app,
@@ -158,12 +165,15 @@ def main():
         _LOGGER.info(f"Running {PKG_NAME} app...")
         bbconf_file_path = os.environ.get("BEDBASE_CONFIG") or None
         global bbagent
-
         global usage_data
-        usage_data = init_model_usage()
+        global ref_validator
+
         bbagent = configure(
             bbconf_file_path
         )  # configure before attaching routers to avoid circular imports
+        usage_data = init_model_usage()
+
+        ref_validator = ReferenceValidator()
 
         scheduler = BackgroundScheduler()
 

bedhost/routers/bed_api.py

@@ -29,15 +29,21 @@
     TokenizedPathResponse,
     QdrantSearchResult,
     RefGenValidReturnModel,
+    RefGenValidModel,
 )
 from fastapi import APIRouter, File, HTTPException, Query, UploadFile, Request
 from fastapi.responses import PlainTextResponse
 from gtars.models import RegionSet
 
 from .. import _LOGGER
 from ..const import EXAMPLE_BED, MAX_FILE_SIZE, MAX_REGION_NUMBER, MIN_REGION_WIDTH
-from ..data_models import CROM_NUMBERS, BaseListResponse, BedDigest
-from ..main import bbagent, usage_data
+from ..data_models import (
+    CROM_NUMBERS,
+    BaseListResponse,
+    BedDigest,
+    ChromLengthUploadModel,
+)
+from ..main import bbagent, usage_data, ref_validator
 from ..helpers import count_requests, test_query_parameter
 
 router = APIRouter(prefix="/v1/bed", tags=["bed"])
@@ -291,6 +297,54 @@ async def embed_bed_file(
     return embedding.tolist()[0]
 
 
+@router.post(
+    "/analyze-genome",
+    summary="Analyze reference genome for bed file",
+    response_model=RefGenValidReturnModel,
+)
+async def analyze_reference_genome(
+    chrom_lengths: ChromLengthUploadModel,
+):
+    """
+    Provide length of the chromosomes for a reference genome, and
+    return reference genome validation results for a bed file
+    """
+
+    try:
+        genome_aliases = bbagent.get_reference_genomes()
+        result = ref_validator.determine_compatibility(
+            chrom_lengths.bed_file, concise=True
+        )
+
+        compared_genomes: List[RefGenValidModel] = []
+        for genome, value in result.items():
+            if value.tier_ranking < 4:
+                compared_genomes.append(
+                    RefGenValidModel(
+                        provided_genome="Not Provided",
+                        compared_genome=genome_aliases.get(genome, "Unknown genome"),
+                        genome_digest=genome,
+                        xs=value.xs,
+                        oobr=value.oobr,
+                        sequence_fit=value.sequence_fit,
+                        assigned_points=value.assigned_points,
+                        tier_ranking=value.tier_ranking,
+                    )
+                )
+        return RefGenValidReturnModel(
+            id="No ID",
+            provided_genome="Not Provided",
+            compared_genome=compared_genomes,
+        )
+
+    except BedBaseConfError as e:
+        _LOGGER.error(e)
+        raise HTTPException(
+            status_code=400,
+            detail=f"Unable to process request. Check loggs",
+        )
+
+
 @router.get(
     "/missing_plots",
     summary="Get missing plots for a bed file.",

requirements/requirements-all.txt

@@ -1,5 +1,5 @@
 # bbconf @ git+https://github.com/databio/bbconf.git@dev#egg=bbconf
-bbconf>=0.14.1
+bbconf>=0.14.2
 fastapi>=0.103.0
 logmuse>=0.2.7
 markdown
@@ -12,4 +12,4 @@ psycopg[binary,pool]
 python-multipart>=0.0.9
 APScheduler>=3.11.0
 cachetools>=4.2.4
-
+bedboss>=0.8.5

test.Dockerfile

@@ -1,7 +1,7 @@
 FROM bedhost
 
-COPY config.yaml /bedbase.yaml
-ENV BEDBASE_CONFIG=/bedbase.yaml
+COPY config.yaml /config.yaml
+ENV BEDBASE_CONFIG=/config.yaml
 
 RUN mkdir -p /data/outputs/bedstat_output
 RUN mkdir -p /data/outputs/bedbuncher_output

ui/package.json

@@ -12,7 +12,7 @@
     "generate-types-local": "npx openapi-typescript http://localhost:8000/openapi.json -o bedbase-types.d.ts"
   },
   "dependencies": {
-    "@databio/gtars": "^0.5.2",
+    "@databio/gtars": "^0.5.3",
     "@tanstack/react-query": "^5.28.0",
     "@tanstack/react-query-devtools": "^5.28.0",
     "@tanstack/react-table": "^8.15.3",