Merge pull request #101 from databio/dev

Release 0.14.3

Author: khoroshevskyi

bbconf/_version.py

@@ -1 +1 @@
-__version__ = "0.14.2"
+__version__ = "0.14.3"

bbconf/bbagent.py

@@ -156,6 +156,14 @@ def get_detailed_stats(self, concise: bool = False) -> FileStats:
                     .order_by(func.count(BedMetadata.assay).desc())
                 ).all()
             }
+            cell_line = {
+                f[0]: f[1]
+                for f in session.execute(
+                    select(BedMetadata.cell_line, func.count(BedMetadata.cell_line))
+                    .group_by(BedMetadata.cell_line)
+                    .order_by(func.count(BedMetadata.cell_line).desc())
+                ).all()
+            }
 
         slice_value = 20
 
@@ -219,12 +227,23 @@ def get_detailed_stats(self, concise: bool = False) -> FileStats:
                 )
                 file_assay_concise.pop("OTHER")
 
+            cell_line_concise = dict(list(cell_line.items())[0:slice_value])
+            cell_line_concise["other"] = sum(
+                list(cell_line.values())[slice_value:]
+            ) + cell_line.get("other", 0)
+            if "" in cell_line_concise:
+                cell_line_concise["other"] = (
+                    cell_line_concise["other"] + cell_line_concise[""]
+                )
+                cell_line_concise.pop("")
+
             return FileStats(
                 data_format=data_format,
                 bed_compliance=bed_compliance_concise,
                 file_genome=file_genomes_concise,
                 file_organism=file_organism_concise,
                 file_assay=file_assay_concise,
+                cell_line=cell_line_concise,
                 bed_comments=bed_comments,
                 geo_status=geo_status,
                 mean_region_width=list_mean_width_bins,
@@ -239,6 +258,7 @@ def get_detailed_stats(self, concise: bool = False) -> FileStats:
             file_genome=file_genomes,
             file_organism=file_organism,
             file_assay=file_assay,
+            cell_line=cell_line,
             bed_comments=bed_comments,
             geo_status=geo_status,
             mean_region_width=list_mean_width_bins,

bbconf/db_utils.py

@@ -532,7 +532,9 @@ class GeoGsmStatus(Base):
         nullable=True, index=True, comment="Bed identifier"
     )
 
-    file_size: Mapped[int] = mapped_column(default=0, comment="Size of the file")
+    file_size: Mapped[int] = mapped_column(
+        BigInteger, default=0, comment="Size of the file"
+    )
     genome: Mapped[str] = mapped_column(nullable=True, comment="Genome")
 
     gse_status_mapper: Mapped["GeoGseStatus"] = relationship(
@@ -727,14 +729,16 @@ def _upload_licenses(self):
         """
         Upload licenses to the database.
         """
+        # Check if licenses already exist to avoid duplicate key errors
+        with Session(self.engine) as session:
+            existing = session.scalar(select(License).limit(1))
+            if existing:
+                _LOGGER.info("Licenses already exist in the database, skipping upload.")
+                return
 
         _LOGGER.info("Uploading licenses to the database...")
         df = pd.read_csv(LICENSES_CSV_URL)
 
-        with Session(self.engine) as session:
-            df.to_sql(
-                License.__tablename__, self.engine, if_exists="append", index=False
-            )
-            session.commit()
+        df.to_sql(License.__tablename__, self.engine, if_exists="append", index=False)
 
         _LOGGER.info("Licenses uploaded successfully!")

bbconf/models/base_models.py

@@ -97,6 +97,7 @@ class FileStats(BaseModel):
     file_genome: Dict[str, int]
     file_organism: Dict[str, int]
     file_assay: Dict[str, int]
+    cell_line: Dict[str, int]
     geo_status: Dict[str, int]
     bed_comments: Dict[str, int]
     mean_region_width: BinValues

docs/changelog.md

@@ -2,6 +2,12 @@
 
 This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html) and [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) format.
 
+### [0.14.3] - 2026-01-31
+### Added:
+- Cell line to detailed bedbase statistics
+
+### Fixed:
+- Saving of big file size (changed to bigint db column type)
 
 ### [0.14.2] - 2026-01-21
 ### Added: