pav3 3.0.0.dev17__tar.gz → 3.0.0.dev18__tar.gz

{pav3-3.0.0.dev17/src/pav3.egg-info → pav3-3.0.0.dev18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pav3
-Version: 3.0.0.dev17
+Version: 3.0.0.dev18
 Summary: Assembly-based variant discovery
 License-Expression: MIT
 Requires-Python: >=3.12

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18}/src/pav3/__init__.py

@@ -1,6 +1,6 @@
 """PAV Python library routines."""
 
-__version__ = '3.0.0.dev17'
+__version__ = '3.0.0.dev18'
 
 __all__ = [
     'align',

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18}/src/pav3/align/trim.py

@@ -83,7 +83,7 @@ TRIM_DESC = {
 }
 """Trim mode description."""
 
-DEFAULT_MIN_TRIM_QRY_LEN = 500
+DEFAULT_MIN_TRIM_QRY_LEN = 1
 """Default minimum query length. Alignments with fewer query bases are removed."""
 
 
@@ -111,7 +111,7 @@ def trim_alignments_qry(
         raise ValueError('Query FAI dataframe is None.')
 
     if min_trim_qry_len < 1:
-        raise ValueError('min_trim_qry_len must be at least 1.')
+        min_trim_qry_len = 1
 
     score_model = get_score_model(score_model)
 

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18}/src/pav3/data/workflow/rules/call.smk

@@ -29,8 +29,6 @@ global temp
 # Rules
 #
 
-CALL_VCF_VARTYPES = ('insdel', 'inv', 'snv', 'cpx', 'dup')
-
 # Generate the VCF
 rule call_vcf:
     input:
@@ -38,7 +36,7 @@ rule call_vcf:
             'results/{asm_name}/call/call_{vartype}.parquet',
             ASM_TABLE, PAV_CONFIG,
             asm_name=wildcards.asm_name,
-            vartype=CALL_VCF_VARTYPES,
+            vartype=pav3.vcf.VARTYPES,
         ),
         callable_ref=lambda wildcards: pav3.pipeline.expand_pattern(
             'results/{asm_name}/call_hap/callable_ref_{hap}.parquet',
@@ -86,14 +84,14 @@ rule call_vcf:
         with pav3.io.TempDirContainer(
                 prefix=f'pav_vcf_{wildcards.asm_name}_'
         ) as temp_file_container:
-            for vartype in CALL_VCF_VARTYPES:
+            for vartype in pav3.vcf.VARTYPES:
 
                 # Read variants
                 df = (
                     pl.scan_parquet(
                         pattern_var.format(
                             asm_name=wildcards.asm_name,
-                            vartype=vartype
+                            vartype=vartype,
                         )
                     )
                     .with_row_index('_index')
@@ -103,11 +101,15 @@ rule call_vcf:
                     df = df.with_columns(pl.lit('SNV').alias('vartype'))
 
                 # Initialize VCF fields (all but FORMAT and sample columns)
-                df = pav3.vcf.init_vcf_fields(
-                    df,
-                    ref_fa=ref_fa,
-                    use_sym=None,
-                )
+                try:
+                    df = pav3.vcf.init_vcf_fields(
+                        df,
+                        ref_fa=ref_fa,
+                        use_sym=None,
+                        vartype=vartype,
+                    )
+                except Exception as e:
+                    raise RuntimeError(f'Failed to initialize VCF fields for {vartype}: {e}') from e
 
                 # Create a sample column
                 df = df.with_columns(
@@ -117,36 +119,45 @@ rule call_vcf:
                 )
 
                 # Append genotypes
-                df = (
-                    df.join(
-                        pav3.vcf.gt_column(
-                            df,
-                            hap_source,
-                            hap_callable,
-                            col_name='_vcf_field_gt',
-                            separator='|',
-                        ),
-                        on='_index',
-                        how='left'
-                    )
-                    .with_columns(
-                        pl.col('_vcf_sample_0').list.concat('_vcf_field_gt'),
-                        pl.col('_vcf_format').list.concat(pl.lit('GT'))
+                try:
+                    df = (
+                        df.join(
+                            pav3.vcf.gt_column(
+                                df,
+                                hap_source,
+                                hap_callable,
+                                col_name='_vcf_field_gt',
+                                separator='|',
+                            ),
+                            on='_index',
+                            how='left'
+                        )
+                        .with_columns(
+                            pl.col('_vcf_sample_0').list.concat('_vcf_field_gt'),
+                            pl.col('_vcf_format').list.concat(pl.lit('GT'))
+                        )
+                        .drop('_vcf_field_gt')
                     )
-                    .drop('_vcf_field_gt')
-                )
+                except Exception as e:
+                    raise RuntimeError(f'Failed to append genotypes for {vartype}: {e}') from e
 
                 # Add INFO fields
-                df = pav3.vcf.standard_info_fields(df)
+                try:
+                    df = pav3.vcf.standard_info_fields(df)
+                except Exception as e:
+                    raise RuntimeError(f'Failed to add INFO fields for {vartype}: {e}') from e
 
                 # Finalize VCF fields
-                df = pav3.vcf.reformat_vcf_table(
-                    df,
-                    sample_columns={0: wildcards.asm_name}
-                )
+                try:
+                    df = pav3.vcf.reformat_vcf_table(
+                        df,
+                        sample_columns={0: wildcards.asm_name}
+                    )
 
-                # Write VCF
-                df.sink_parquet(temp_file_container.next())
+                    # Write VCF
+                    df.sink_parquet(temp_file_container.next())
+                except Exception as e:
+                    raise RuntimeError(f'Failed to finalize VCF fields for {vartype}: {e}') from e
 
             # Write
             header_list = pav3.vcf.get_headers(

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18}/src/pav3/lgsv/struct.py

@@ -143,6 +143,9 @@ def get_ref_trace(
                     on='_index_aligned',
                     how='left'
                 )
+                .with_columns(
+                    pl.col('is_rev') ^ interval.is_rev
+                )
                 .group_by('_index_trace')
                 .agg(
                     (
@@ -159,6 +162,13 @@ def get_ref_trace(
         )
         .drop(['_index_aligned', '_index_trace'])
 
+        # Adjust anchor-only segments at the edges to include only segments that are a single fwd copy
+        .with_columns(
+            (pl.col('_all_anchor') & (
+                (pl.col('fwd_count') == 1) & (pl.col('rev_count') == 0)
+            )).alias('_all_anchor')
+        )
+
         # Drop anchor-only segments at the edges
         .filter(
             ~pl.col('_all_anchor') | (

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18}/src/pav3/vcf.py

@@ -1,7 +1,30 @@
 """Transform variant calls from Parquet files to VCF files."""
 
+__all__ = [
+    'InfoField',
+    'FilterField',
+    'FormatField',
+    'AltField',
+    'VCF_VERSION',
+    'INFO_FIELDS',
+    'FILTER_FIELDS',
+    'FORMAT_FIELDS',
+    'ALT_FIELDS',
+    'VARTYPES',
+    'VARTYPE_COMPAT',
+    'get_headers',
+    'init_vcf_fields',
+    'vcf_fields_seq_insdel',
+    'vcf_fields_sym',
+    'vcf_fields_snv',
+    'gt_column',
+    'standard_info_fields',
+    'reformat_vcf_table',
+    'get_hap_source',
+]
+
+
 from collections.abc import Iterable, Mapping
-from dataclasses import dataclass
 import datetime
 from pathlib import Path
 import threading
@@ -170,6 +193,18 @@ ALT_FIELDS = [
     AltField('CPX', 'Complex variant'),
 ]
 
+VARTYPES = ('insdel', 'inv', 'snv', 'cpx', 'dup')
+
+VARTYPE_COMPAT = {
+    'insdel': {'INS', 'DEL'},
+    'ins': {'INS'},
+    'del': {'DEL'},
+    'inv': {'INV'},
+    'snv': {'SNV'},
+    'cpx': {'CPX'},
+    'dup': {'DUP'},
+}
+
 
 def get_headers(
         ref_filename: Optional[str | Path] = None,
@@ -221,7 +256,7 @@ def get_headers(
 
 def init_vcf_fields(
         df: pl.LazyFrame,
-        ref_fa: Optional[str] = None,
+        ref_fa: Optional[str | Path] = None,
         use_sym: Optional[bool] = None,
         vartype: Optional[str] = None,
 ) -> pl.LazyFrame:
@@ -234,11 +269,30 @@ def init_vcf_fields(
     :param ref_fa: Reference FASTA filename. Required for non-SNV variants.
     :param use_sym: Whether to use symbolic alternate records. If None, INS, DEL, and SNV ore not
         symbolic, and all other variant types are.
+    :param vartype: Expected variant type or types (i.e. "insdel" allows "INS" and "DEL"). Setting
+        this field ensures the variant types is handled consisntently if all variant types
+        are not in table `df` or `df` is empty.
     """
+    df_schema = df.collect_schema()
+
     vartype_set = set(
         df.select(pl.col('vartype').unique()).collect().to_series()
     )
 
+    if vartype is not None:
+        vartype_compat = VARTYPE_COMPAT.get(vartype, None)
+
+        if vartype_compat is None:
+            raise ValueError(f'Unknown variant type: {vartype}')
+
+        if no_compat_set := vartype_set - vartype_compat:
+            raise ValueError(
+                f'Declared variant type "{vartype}" is not compatible with types found '
+                f'in the variant table: {", ".join(sorted([str(_).upper() for _ in no_compat_set]))}'
+            )
+
+        vartype_set = vartype_compat
+
     if 'SNV' in vartype_set and len(vartype_set) > 1:
         raise ValueError(
             f'Cannot mix SNV variants with other variant types: '
@@ -335,7 +389,7 @@ def init_vcf_fields(
 
 def vcf_fields_seq_insdel(
         df: pl.LazyFrame,
-        ref_fa: str,
+        ref_fa: Path | str,
 ) -> pl.LazyFrame:
     """Set POS, REF, and ALT VCF fields for insertions and deletions (non-symbolic).
 
@@ -464,7 +518,7 @@ def vcf_fields_seq_insdel(
 
 def vcf_fields_sym(
     df: pl.LazyFrame,
-    ref_fa: str,
+    ref_fa: Path | str,
 ) -> pl.LazyFrame:
     """Get POS, REF, and ALT fields for symbolic alternate records.
 
@@ -955,17 +1009,17 @@ class _LockingPysam():
 
     Pysam is not thread-safe, and will fail when called through Polars. This class explicitly locks it.
     """
-    filename: str
+    filename: Path
 
     def __init__(
             self,
-            filename,
+            filename: Path | str,
     ):
         """Create a new locking object.
 
         :param filename: FASTA file name.
         """
-        self.filename = filename
+        self.filename = Path(filename)
         self.ref_file = None
         self.lock = threading.Lock()
 
@@ -973,7 +1027,7 @@ class _LockingPysam():
         if self.ref_file is not None:
             raise RuntimeError('Ref file already open')
 
-        self.ref_file = pysam.FastaFile(self.filename)
+        self.ref_file = pysam.FastaFile(str(self.filename))
 
         return self
 
@@ -995,124 +1049,3 @@ class _LockingPysam():
 
         with self.lock:
             return self.ref_file.fetch(chrom, start, end)
-
-
-#
-# Refactor
-#
-
-
-# def _init_vcf_fields(
-#         df: pl.LazyFrame | pl.DataFrame,
-# ) -> pl.LazyFrame:
-#     """Initialize VCF fields common across all variant types (does not set REF, ALT, or POS).
-#
-#     :param df: Variant table.
-#     """
-#
-#     if isinstance(df, pl.DataFrame):
-#         df = df.lazy()
-#
-#     return (
-#         df.with_columns(
-#             pl.col('chrom').alias('#CHROM'),
-#             pl.col('id').alias('ID'),
-#             pl.lit('.').alias('QUAL'),
-#             (
-#                 pl.when(pl.col('filter').is_not_null())
-#                 .then(pl.col('filter').list.join(';'))
-#                 .otherwise(pl.lit('PASS'))
-#             ).alias('FILTER'),
-#             pl.lit([]).cast(pl.List(pl.String)).alias('_info'),
-#             pl.lit([]).cast(pl.List(pl.Struct({'fmt': pl.String, 'sample': pl.String}))).alias('_sample')
-#         )
-#         .with_columns(
-#             pl.col('_info').list.concat([
-#                 pl.concat_str(pl.lit('ID='), 'id'),
-#                 pl.concat_str(pl.lit('QRY_REGION='), pl.col('chrom'), pl.lit(':'), pl.col('qry_pos') + 1, pl.lit('-'), pl.col('qry_end')),
-#                 # pl.concat_str(pl.lit('QRY_STRAND='), pl.when(pl.col('is_rev')).then('-').otherwise('+')),
-#                 pl.concat_str(pl.lit('ALIGN_INDEX='), pl.col('align_index').cast(pl.List(pl.String)).list.join(','))
-#             ])
-#         )
-#     )
-
-# def _init_table():
-#     df = (
-#         df.with_columns(
-#             pl.col('chrom').alias('#CHROM'),
-#             # POS
-#             pl.col('id').alias('ID'),
-#             # REF
-#             # ALT
-#             pl.lit('.').alias('QUAL'),
-#             (
-#                 pl.when(pl.col('filter').is_not_null())
-#                 .then(pl.col('filter').list.join(';'))
-#                 .otherwise(pl.lit('PASS'))
-#             ).alias('FILTER'),
-#             # INFO
-#             # FORMAT
-#             # Sample
-#         )
-#     )
-
-# def _init_table_other(
-#
-# ):
-#     df = (
-#         df.with_columns(
-#             (pl.col('vartype').is_in(['INV', 'CPX', 'DUP'])).alias('_is_sym'),
-#             (pl.col('vartype') == 'SNV').alias('_is_snv'),
-#             (pl.col('pos') > 0).alias('_ref_left')  # Append reference base to left if True (most variants), move to end if not (variant at position 0)
-#         )
-#     )
-
-#
-# def _ref_base(df, ref_fa):
-#     """
-#     Get reference base preceding a variant (SV, indel) or at the point change (SNV).
-#
-#     :param df: Variant dataframe as BED.
-#     :param ref_fa: Reference file.
-#     """
-#
-#     # Get the reference base location
-#     with pysam.FastaFile(ref_fa) as ref_file:
-#
-#         df = (
-#             df.with_columns(
-#                 (
-#                     pl.when(pl.col('_ref_left'))
-#                     .then(pl.col('pos') - 1)
-#                     .otherwise(pl.col('end'))
-#                 ).alias('_ref_base_loc')
-#             )
-#             .with_columns(
-#                 pl.when(pl.col('_is_snv'))
-#                 .then(pl.col('_ref_base_loc') + 1)
-#                 .otherwise(pl.col('_ref_base_loc'))
-#             ).alias('_ref_base_loc')
-#             .with_columns(
-#                 pl.struct(['chrom', '_ref_base_loc'])
-#                 .apply(
-#                     lambda vals: ref_file.fetch(vals['chrom'], vals['_ref_base_loc'], vals['_ref_base_loc'] + 1)
-#                 ).alias('_ref_base')
-#             )
-#             .drop('_ref_base_Loc')
-#         )
-#
-#     # Open and update records
-#     with pysam.FastaFile(ref_fa) as ref_file:
-#         for index, row in df.iterrows():
-#
-#             if row['SVTYPE'] in {'INS', 'DEL', 'INSDEL', 'DUP', 'INV'}:
-#                 yield ref_file.fetch(row['#CHROM'], row['POS'] + (-1 if row['POS'] > 0 else 0), row['POS']).upper()
-#
-#             elif row['SVTYPE'] == 'SNV':
-#                 if 'REF' in row:
-#                     yield row['REF']
-#                 else:
-#                     yield ref_file.fetch(row['#CHROM'], row['POS'], row['POS'] + 1).upper()
-#
-#             else:
-#                 raise RuntimeError('Unknown variant type: "{}" at index {}'.format(row['VARTYPE'], index))

{pav3-3.0.0.dev17 → pav3-3.0.0.dev18/src/pav3.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pav3
-Version: 3.0.0.dev17
+Version: 3.0.0.dev18
 Summary: Assembly-based variant discovery
 License-Expression: MIT
 Requires-Python: >=3.12