pav3 3.0.0.dev15__tar.gz → 3.0.0.dev17__tar.gz

{pav3-3.0.0.dev15/src/pav3.egg-info → pav3-3.0.0.dev17}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pav3
-Version: 3.0.0.dev15
+Version: 3.0.0.dev17
 Summary: Assembly-based variant discovery
 License-Expression: MIT
 Requires-Python: >=3.12

{pav3-3.0.0.dev15 → pav3-3.0.0.dev17}/src/pav3/__init__.py

@@ -1,6 +1,6 @@
 """PAV Python library routines."""
 
-__version__ = '3.0.0.dev15'
+__version__ = '3.0.0.dev17'
 
 __all__ = [
     'align',

{pav3-3.0.0.dev15 → pav3-3.0.0.dev17}/src/pav3/call/intra.py

@@ -21,6 +21,7 @@ __all__ = [
 
 import agglovar
 import os
+from pathlib import Path
 from typing import Optional
 
 import Bio.Seq
@@ -363,8 +364,8 @@ def variant_tables_snv_insdel(
 def variant_tables_inv(
         df_align: pl.DataFrame | pl.LazyFrame,
         df_flag: pl.DataFrame | pl.LazyFrame,
-        ref_fa_filename: str,
-        qry_fa_filename: str,
+        ref_fa_filename: str | Path,
+        qry_fa_filename: str | Path,
         df_ref_fai: pl.DataFrame,
         df_qry_fai: pl.DataFrame,
         pav_params: Optional[PavParams] = None,
@@ -408,8 +409,12 @@ def variant_tables_inv(
     )
 
     # Create variant tables
+    inv_schema = {col: type_ for col, type_ in schema.VARIANT.items() if col in set(get_inv_row().keys())}
+
     variant_table_list = []
 
+    log_file = None
+
     for row in df_flag.iter_rows(named=True):
         region_flag = Region(
             chrom=row['chrom'], pos=row['pos'], end=row['end'],
@@ -427,23 +432,17 @@ def variant_tables_inv(
             k_util=k_util,
             kde_model=kde_model,
             stop_on_lift_fail=True,
-            log_file=None,
+            log_file=log_file,
         )
 
         if inv_row is not None:
-            inv_row['align_index'] = [row['align_index']]
-
-            variant_table_list.append(pl.DataFrame(inv_row))
-
-    col_set = set(get_inv_row().keys())
-
-    col_set.add('filter')
-    col_set.add('align_source')
-
-    table_schema = {col: type_ for col, type_ in schema.VARIANT.items() if col in col_set}
+            variant_table_list.append(inv_row)
 
     df_inv = (
-        pl.from_dicts(variant_table_list, schema=table_schema)
+        pl.from_dicts(
+            variant_table_list,
+            schema=inv_schema,
+        )
         .with_columns(
             expr.id_nonsnv().alias('id'),
             pl.lit(CALL_SOURCE).alias('call_source')
@@ -454,7 +453,6 @@ def variant_tables_inv(
             right_on='align_index',
             how='left'
         )
-        .select(list(table_schema.keys()))
     )
 
     return schema.cast(df_inv, schema.VARIANT)

{pav3-3.0.0.dev15 → pav3-3.0.0.dev17}/src/pav3/data/workflow/rules/call.smk

@@ -862,8 +862,8 @@ rule call_intra_inv:
         df_inv = pav3.call.intra.variant_tables_inv(
             df_align=df_align,
             df_flag=df_flag,
-            ref_fa_filename=str(ref_fa_filename),
-            qry_fa_filename=str(qry_fa_filename),
+            ref_fa_filename=ref_fa_filename,
+            qry_fa_filename=qry_fa_filename,
             df_ref_fai=df_ref_fai,
             df_qry_fai=df_qry_fai,
             pav_params=pav_params,

{pav3-3.0.0.dev15 → pav3-3.0.0.dev17}/src/pav3/schema.py

@@ -168,6 +168,7 @@ Fields:
 def cast(
     df: pl.DataFrame,
     schema: dict[str, pl.DataType],
+    do_sort: bool = True,
 ) -> pl.DataFrame:
     ...
 
@@ -175,22 +176,48 @@ def cast(
 def cast(
     df: pl.LazyFrame,
     schema: dict[str, pl.DataType],
+    do_sort: bool = True,
 ) -> pl.LazyFrame:
     ...
 
 def cast(
     df: pl.DataFrame | pl.LazyFrame,
     schema: dict[str, pl.DataType],
+    do_sort: bool = True,
 ) -> pl.DataFrame | pl.LazyFrame:
+    """
+    Cast a DataFrame or LazyFrame to a schema.
+
+    :param df: DataFrame or LazyFrame to cast.
+    :param schema: Schema to cast to.
+    :param do_sort: Whether to sort the DataFrame or LazyFrame by the order defined in the schema.
+        If the table contains columns not defined by the schema, they are retained in order after
+        columns defined by the schema.
+    """
 
     if isinstance(df, pl.DataFrame):
         existing_schema = df.schema
     else:
         existing_schema = df.collect_schema()
 
-    return df.cast(
+    df = df.cast(
         {
             col: schema.get(col, type_)
             for col, type_ in existing_schema.items()
         }
     )
+
+    if do_sort:
+        df = (
+            df
+            .select(
+                *[  # Known columns first
+                    col for col in schema.keys() if col in existing_schema.keys()
+                ],
+                *[  # Unknown columns last
+                    col for col in existing_schema.keys() if col not in schema.keys()
+                ]
+            )
+        )
+
+    return df

{pav3-3.0.0.dev15 → pav3-3.0.0.dev17/src/pav3.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pav3
-Version: 3.0.0.dev15
+Version: 3.0.0.dev17
 Summary: Assembly-based variant discovery
 License-Expression: MIT
 Requires-Python: >=3.12