biofiles 0.0.6__py3-none-any.whl → 0.0.7__py3-none-any.whl

biofiles/feature.py

@@ -1,10 +1,10 @@
 from collections import deque
 from dataclasses import dataclass, field
 from pathlib import Path
-from typing import Iterator, TextIO
+from typing import Iterator, TextIO, Type
 
 from biofiles.common import Reader, Strand
-from biofiles.types.feature import Feature, Gene, ThreePrimeUTR, Exon
+from biofiles.types.feature import Feature, Gene, ThreePrimeUTR, Exon, UTR
 
 
 @dataclass
@@ -67,6 +67,9 @@ class FeatureReader(Reader):
         super().__init__(input_)
         self._streaming_window = streaming_window
 
+    def __iter__(self) -> Iterator[Feature]:
+        raise NotImplementedError
+
     def _finalize_drafts(
         self, drafts: FeatureDrafts, w: int | None
     ) -> Iterator[Feature]:
@@ -109,7 +112,9 @@ class FeatureReader(Reader):
             case "exon":
                 feature = self._finalize_exon(draft, result)
             case "three_prime_utr":
-                feature = self._finalize_three_prime_utr(draft, result)
+                feature = self._finalize_utr(draft, result, ThreePrimeUTR)
+            case "utr":
+                feature = self._finalize_utr(draft, result, UTR)
             case _:
                 feature = self._finalize_other(draft, result)
         if feature.parent:
@@ -138,8 +143,8 @@ class FeatureReader(Reader):
         object.__setattr__(gene, "exons", gene.exons + (exon,))
         return exon
 
-    def _finalize_three_prime_utr(
-        self, draft: FeatureDraft, result: Features
+    def _finalize_utr(
+        self, draft: FeatureDraft, result: Features, type_: Type[UTR]
     ) -> Feature:
         feature = self._finalize_other(draft, result)
 
@@ -149,7 +154,7 @@ class FeatureReader(Reader):
 
         if gene is None:
             return feature
-        return ThreePrimeUTR(**feature.__dict__, gene=gene)
+        return type_(**feature.__dict__, gene=gene)
 
     def _finalize_other(self, draft: FeatureDraft, result: Features) -> Feature:
         parent_id = self._extract_parent_id(draft)

biofiles/gff.py

@@ -122,8 +122,8 @@ class GFF3Writer(Writer):
             feature.sequence_id,
             feature.source,
             feature.type_,
-            str(feature.start_original),
-            str(feature.end_original),
+            str(feature.start_c + 1),
+            str(feature.end_c),
             str(feature.score) if feature.score is not None else ".",
             str(feature.strand) if feature.strand is not None else ".",
             str(feature.phase) if feature.phase is not None else ".",

biofiles/types/feature.py

@@ -14,7 +14,7 @@ class Feature:
 
     start_original: int
     end_original: int
-    # Original, 1-based inclusive values.
+    # Original values as they were present in the file (1-based inclusive for .gff and .gtf).
 
     start_c: int
     end_c: int
@@ -47,6 +47,11 @@ class Exon(Feature):
 
 
 @dataclass(frozen=True)
-class ThreePrimeUTR(Feature):
+class UTR(Feature):
     gene: Gene
     # TODO transcript
+
+
+@dataclass(frozen=True)
+class ThreePrimeUTR(UTR):
+    pass

{biofiles-0.0.6.dist-info → biofiles-0.0.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: biofiles
-Version: 0.0.6
+Version: 0.0.7
 Summary: Pure-Python, zero-dependency collection of bioinformatics-related file readers and writers
 Author-email: Tigran Saluev <tigran@saluev.com>
 Maintainer-email: Tigran Saluev <tigran@saluev.com>

{biofiles-0.0.6.dist-info → biofiles-0.0.7.dist-info}/RECORD

@@ -1,16 +1,16 @@
 biofiles/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 biofiles/common.py,sha256=Yi0i85FpD2wR3vqL645LTUAE6TybGDxxZQsUmEGHqu4,1126
 biofiles/fasta.py,sha256=ctIt5I_fcZx-xQN921zpmlZS7e9_ICf-3_i6mTs5qbs,2135
-biofiles/feature.py,sha256=4eRXmusxTbrDqkwPkMKaa1nYRSSYvt3Y3_-7jC7rkII,6693
-biofiles/gff.py,sha256=TwBasmakNmeS5yikP-E5iMEvB6BXgczGYQKiGa2_LAw,5516
+biofiles/feature.py,sha256=wga91SYc2cMh5r5nH2m9IRM5hnVObesrxYr31kBU4WA,6874
+biofiles/gff.py,sha256=W5AjaQL_iYk4OF-H7C2pOjtpeLDEKfVg5uTOFxPDJ5I,5506
 biofiles/gtf.py,sha256=hWfjQjzwsrXLjCGr9ia6GdHNdYtlwkBrG1ldJYhRD-4,1251
 biofiles/repeatmasker.py,sha256=DqD1z1hUfCP4-qnfjF-oMF-ZpW_6XhOf_nzA8VHhQbw,3079
 biofiles/types/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-biofiles/types/feature.py,sha256=PyxIkbpWLcboY8cGEcJBn6sWv5wIFgOyD9Ey5D2G3JQ,938
+biofiles/types/feature.py,sha256=gdywZKp7_TXiiZcXfbVs6oSHIWtju2-bqfQOmjH71Dg,1041
 biofiles/types/repeat.py,sha256=63SqzAwEGIDIGP9pxC85RUdwXbbSm0S5WNL3lSiWlmc,641
 biofiles/types/sequence.py,sha256=EOw_oKuMR0THpCYJqVE__27z7qrRqcdIPrRWTL4OFMw,152
-biofiles-0.0.6.dist-info/LICENSE,sha256=CbR8ssdFyViKj25JAlMjIt1_FbiZ1tAC5t-uwUbxqak,1070
-biofiles-0.0.6.dist-info/METADATA,sha256=vbIWtDF_yHm-qV8EKlCiKQzjajnoIZdV0RWZJoS0zh0,3033
-biofiles-0.0.6.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-biofiles-0.0.6.dist-info/top_level.txt,sha256=laFaFv8hpkI4U-Pgs0yBaAJXN2_CJKl7jb-m3-tGfSc,9
-biofiles-0.0.6.dist-info/RECORD,,
+biofiles-0.0.7.dist-info/LICENSE,sha256=CbR8ssdFyViKj25JAlMjIt1_FbiZ1tAC5t-uwUbxqak,1070
+biofiles-0.0.7.dist-info/METADATA,sha256=jd6-bM7tz1laNRX_wUCnDP4FRvQwGO0sxqqWYAvowiQ,3033
+biofiles-0.0.7.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+biofiles-0.0.7.dist-info/top_level.txt,sha256=laFaFv8hpkI4U-Pgs0yBaAJXN2_CJKl7jb-m3-tGfSc,9
+biofiles-0.0.7.dist-info/RECORD,,