cool-seq-tool 0.13.1__py3-none-any.whl → 0.14.0__py3-none-any.whl

cool_seq_tool/mappers/exon_genomic_coords.py

@@ -93,6 +93,9 @@ class GenomicTxSeg(BaseModelForbidExtra):
     )
     genomic_ac: StrictStr | None = Field(None, description="RefSeq genomic accession.")
     tx_ac: StrictStr | None = Field(None, description="RefSeq transcript accession.")
+    strand: Strand | None = Field(
+        None, description="The strand that the transcript accession exists on."
+    )
     errors: list[StrictStr] = Field([], description="Error messages.")
 
     @model_validator(mode="before")
@@ -121,6 +124,7 @@ class GenomicTxSeg(BaseModelForbidExtra):
                 "gene": "TPM3",
                 "genomic_ac": "NC_000001.11",
                 "tx_ac": "NM_152263.3",
+                "strand": -1,
                 "seg": {
                     "exon_ord": 0,
                     "offset": 0,
@@ -147,6 +151,9 @@ class GenomicTxSegService(BaseModelForbidExtra):
     )
     genomic_ac: StrictStr | None = Field(None, description="RefSeq genomic accession.")
     tx_ac: StrictStr | None = Field(None, description="RefSeq transcript accession.")
+    strand: Strand | None = Field(
+        None, description="The strand that the transcript exists on."
+    )
     seg_start: TxSegment | None = Field(None, description="Start transcript segment.")
     seg_end: TxSegment | None = Field(None, description="End transcript segment.")
     errors: list[StrictStr] = Field([], description="Error messages.")
@@ -183,6 +190,7 @@ class GenomicTxSegService(BaseModelForbidExtra):
                 "gene": "TPM3",
                 "genomic_ac": "NC_000001.11",
                 "tx_ac": "NM_152263.3",
+                "strand": -1,
                 "seg_start": {
                     "exon_ord": 0,
                     "offset": 0,
@@ -400,6 +408,7 @@ class ExonGenomicCoordsMapper:
             gene=gene,
             genomic_ac=genomic_ac,
             tx_ac=transcript,
+            strand=strand,
             seg_start=seg_start,
             seg_end=seg_end,
         )
@@ -490,6 +499,7 @@ class ExonGenomicCoordsMapper:
             params["gene"] = start_tx_seg_data.gene
             params["genomic_ac"] = start_tx_seg_data.genomic_ac
             params["tx_ac"] = start_tx_seg_data.tx_ac
+            params["strand"] = start_tx_seg_data.strand
             params["seg_start"] = start_tx_seg_data.seg
         else:
             start_tx_seg_data = None
@@ -524,6 +534,7 @@ class ExonGenomicCoordsMapper:
                 params["gene"] = end_tx_seg_data.gene
                 params["genomic_ac"] = end_tx_seg_data.genomic_ac
                 params["tx_ac"] = end_tx_seg_data.tx_ac
+                params["strand"] = end_tx_seg_data.strand
 
             params["seg_end"] = end_tx_seg_data.seg
 
@@ -865,16 +876,21 @@ class ExonGenomicCoordsMapper:
         if use_alt_start_i and coordinate_type == CoordinateType.RESIDUE:
             genomic_pos = genomic_pos - 1  # Convert residue coordinate to inter-residue
 
-        # Validate that the breakpoint between the first and last exon for the selected transcript
+        # Validate that the breakpoint occurs within 150 bp of the first and last exon for the selected transcript.
+        # A breakpoint beyond this range is likely erroneous.
         coordinate_check = await self._validate_genomic_breakpoint(
             pos=genomic_pos, genomic_ac=genomic_ac, tx_ac=transcript
         )
         if not coordinate_check:
-            return GenomicTxSeg(
-                errors=[
-                    f"{genomic_pos} on {genomic_ac} does not occur within the exons for {transcript}"
-                ]
+            msg = (
+                f"{genomic_pos} on {genomic_ac} occurs more than 150 bp outside the "
+                f"exon boundaries of the {transcript} transcript, indicating this may not "
+                f"be a chimeric transcript junction and is unlikely to represent a "
+                f"contiguous coding sequence. Confirm that the genomic position "
+                f"{genomic_pos} is being used to represent transcript junction and not "
+                f"DNA breakpoint."
             )
+            _logger.warning(msg)
 
         # Check if breakpoint occurs on an exon.
         # If not, determine the adjacent exon given the selected transcript
@@ -913,6 +929,7 @@ class ExonGenomicCoordsMapper:
             gene=gene,
             genomic_ac=genomic_ac,
             tx_ac=transcript,
+            strand=strand,
             seg=TxSegment(
                 exon_ord=exon_num,
                 offset=offset,
@@ -955,8 +972,9 @@ class ExonGenomicCoordsMapper:
         :param pos: Genomic position on ``genomic_ac``
         :param genomic_ac: RefSeq genomic accession, e.g. ``"NC_000007.14"``
         :param transcript: A transcript accession
-        :return: ``True`` if the coordinate falls within the first and last exon
-            for the transcript, ``False`` if not
+        :return: ``True`` if the coordinate falls within 150bp of the first and last exon
+            for the transcript, ``False`` if not. Breakpoints past this threshold
+            are likely erroneous.
         """
         query = f"""
             WITH tx_boundaries AS (
@@ -968,7 +986,7 @@ class ExonGenomicCoordsMapper:
                 AND alt_ac = '{genomic_ac}'
             )
             SELECT * FROM tx_boundaries
-            WHERE {pos} between tx_boundaries.min_start and tx_boundaries.max_end
+            WHERE {pos} between (tx_boundaries.min_start - 150) and (tx_boundaries.max_end + 150)
             """  # noqa: S608
         results = await self.uta_db.execute_query(query)
         return bool(results)

{cool_seq_tool-0.13.1.dist-info → cool_seq_tool-0.14.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cool_seq_tool
-Version: 0.13.1
+Version: 0.14.0
 Summary: Common Operation on Lots of Sequences Tool
 Author: Kori Kuzma, James Stevenson, Katie Stahl, Alex Wagner
 License: MIT License

{cool_seq_tool-0.13.1.dist-info → cool_seq_tool-0.14.0.dist-info}/RECORD

@@ -6,7 +6,7 @@ cool_seq_tool/handlers/__init__.py,sha256=KalQ46vX1MO4SJz2SlspKoIRy1n3c3Vp1t4Y2p
 cool_seq_tool/handlers/seqrepo_access.py,sha256=Jd19jbdUvPRPn_XWozL67ph-nSIxpb4_UUimapDrsm4,9162
 cool_seq_tool/mappers/__init__.py,sha256=4_YNwNyw_QrlhRNu1nly8Dezv81XjCIiNa7crVXEh38,305
 cool_seq_tool/mappers/alignment.py,sha256=nV6PS3mhkQ2MD1GcpNBujBOqd3AKxYSYA9BCusFOa1o,9636
-cool_seq_tool/mappers/exon_genomic_coords.py,sha256=MPp1cMfaTYZRJ-T9cP8SvICJbTlTCL5Yze3J1VT-oQQ,43790
+cool_seq_tool/mappers/exon_genomic_coords.py,sha256=uPBlI9K_IUV3xN9unnCwd6bPmqpfIWB5I-dksN_JZSw,44765
 cool_seq_tool/mappers/liftover.py,sha256=lltx9zxfkrb5PHtJlKp3a39JCwPP4e0Zft-mQc1jXL8,3367
 cool_seq_tool/mappers/mane_transcript.py,sha256=C9eKEj8qhVg878oUhBKPYAZS7gpLM5aaQ0HhSkUg-2g,54365
 cool_seq_tool/resources/__init__.py,sha256=VwUC8YaucTS6SmRirToulZTF6CuvuLQRSxFfSfAovCc,77
@@ -17,8 +17,8 @@ cool_seq_tool/sources/__init__.py,sha256=51QiymeptF7AeVGgV-tW_9f4pIUr0xtYbyzpvHO
 cool_seq_tool/sources/mane_transcript_mappings.py,sha256=C5puIA1xuEzBaSvs8VtSxVb2OIDGUg5no8v6Ma2QSdw,6597
 cool_seq_tool/sources/transcript_mappings.py,sha256=903RKTMBO2rbKh6iTQ1BEWnY4C7saBFMPw2_4ATuudg,10054
 cool_seq_tool/sources/uta_database.py,sha256=UHFLeiuk8H29CF1tNjE8T22-QaPs_fDUaqQO6Hu18yg,36175
-cool_seq_tool-0.13.1.dist-info/licenses/LICENSE,sha256=IpqC9A-tZW7XXXvCS8c4AVINqkmpxiVA-34Qe3CZSjo,1072
-cool_seq_tool-0.13.1.dist-info/METADATA,sha256=ZhMjXeb_uT0a9hITKtjhUzVDJtlCJKLLPpOf2Cg8GpI,6579
-cool_seq_tool-0.13.1.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
-cool_seq_tool-0.13.1.dist-info/top_level.txt,sha256=cGuxdN6p3y16jQf6hCwWhE4OptwUeZPm_PNJlPb3b0k,14
-cool_seq_tool-0.13.1.dist-info/RECORD,,
+cool_seq_tool-0.14.0.dist-info/licenses/LICENSE,sha256=IpqC9A-tZW7XXXvCS8c4AVINqkmpxiVA-34Qe3CZSjo,1072
+cool_seq_tool-0.14.0.dist-info/METADATA,sha256=1BpHtcCsaCUBvpmp1Qe1TVoZFHVvUir685EsH925PJ0,6579
+cool_seq_tool-0.14.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+cool_seq_tool-0.14.0.dist-info/top_level.txt,sha256=cGuxdN6p3y16jQf6hCwWhE4OptwUeZPm_PNJlPb3b0k,14
+cool_seq_tool-0.14.0.dist-info/RECORD,,

{cool_seq_tool-0.13.1.dist-info → cool_seq_tool-0.14.0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (79.0.0)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any