cool-seq-tool 0.13.1__tar.gz → 0.14.0__tar.gz

{cool_seq_tool-0.13.1 → cool_seq_tool-0.14.0}/PKG-INFO

@@ -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 → cool_seq_tool-0.14.0}/src/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 → cool_seq_tool-0.14.0}/src/cool_seq_tool.egg-info/PKG-INFO

@@ -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 → cool_seq_tool-0.14.0}/tests/mappers/test_exon_genomic_coords.py

@@ -1,5 +1,6 @@
 """Module for testing that Cool Seq Tool works correctly."""
 
+import logging
 from datetime import datetime
 
 import pytest
@@ -171,6 +172,7 @@ def tpm3_exon1():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "strand": -1,
         "seg": {
             "exon_ord": 0,
             "offset": 0,
@@ -194,6 +196,7 @@ def tpm3_exon8():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "strand": -1,
         "seg": {
             "exon_ord": 7,
             "offset": 0,
@@ -217,6 +220,7 @@ def tpm3_exon1_g(tpm3_exon1):
         "gene": tpm3_exon1.gene,
         "genomic_ac": tpm3_exon1.genomic_ac,
         "tx_ac": tpm3_exon1.tx_ac,
+        "strand": tpm3_exon1.strand,
         "seg_start": tpm3_exon1.seg,
     }
     return GenomicTxSegService(**params)
@@ -229,6 +233,7 @@ def tpm3_exon8_g(tpm3_exon8):
         "gene": tpm3_exon8.gene,
         "genomic_ac": tpm3_exon8.genomic_ac,
         "tx_ac": tpm3_exon8.tx_ac,
+        "strand": tpm3_exon8.strand,
         "seg_end": tpm3_exon8.seg,
     }
     return GenomicTxSegService(**params)
@@ -241,6 +246,7 @@ def tpm3_exon1_exon8(tpm3_exon1, tpm3_exon8):
         "gene": tpm3_exon8.gene,
         "genomic_ac": tpm3_exon8.genomic_ac,
         "tx_ac": tpm3_exon8.tx_ac,
+        "strand": tpm3_exon8.strand,
         "seg_start": tpm3_exon1.seg,
         "seg_end": tpm3_exon8.seg,
     }
@@ -261,6 +267,7 @@ def tpm3_exon1_exon8_offset(tpm3_exon1, tpm3_exon8):
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "strand": -1,
         "seg_start": tpm3_exon1_cpy.seg,
         "seg_end": tpm3_exon8_cpy.seg,
     }
@@ -274,6 +281,7 @@ def mane_braf():
         "gene": "BRAF",
         "genomic_ac": "NC_000007.14",
         "tx_ac": "NM_004333.6",
+        "strand": -1,
         "seg_start": {
             "exon_ord": 5,
             "offset": 1,
@@ -309,6 +317,7 @@ def wee1_exon2_exon11():
         "gene": "WEE1",
         "genomic_ac": "NC_000011.10",
         "tx_ac": "NM_003390.3",
+        "strand": 1,
         "seg_start": {
             "exon_ord": 1,
             "offset": 205,
@@ -344,6 +353,7 @@ def mane_wee1_exon2_exon11():
         "gene": "WEE1",
         "genomic_ac": "NC_000011.10",
         "tx_ac": "NM_003390.4",
+        "strand": 1,
         "seg_start": {
             "exon_ord": 1,
             "offset": 205,
@@ -379,6 +389,7 @@ def ntrk1_exon10_exon17():
         "gene": "NTRK1",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_002529.3",
+        "strand": 1,
         "seg_start": {
             "exon_ord": 9,
             "offset": 0,
@@ -414,6 +425,7 @@ def zbtb10_exon3_end():
         "gene": "ZBTB10",
         "genomic_ac": "NC_000008.11",
         "tx_ac": "NM_001105539.3",
+        "strand": 1,
         "seg_start": None,
         "seg_end": {
             "exon_ord": 2,
@@ -438,6 +450,7 @@ def zbtb10_exon5_start():
         "gene": "ZBTB10",
         "genomic_ac": "NC_000008.11",
         "tx_ac": "NM_001105539.3",
+        "strand": 1,
         "seg_start": {
             "exon_ord": 4,
             "offset": -201,
@@ -462,6 +475,7 @@ def tpm3_exon6_end():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.4",
+        "strand": -1,
         "seg_start": None,
         "seg_end": {
             "exon_ord": 5,
@@ -486,6 +500,7 @@ def tpm3_exon5_start():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.4",
+        "strand": -1,
         "seg_start": {
             "exon_ord": 4,
             "offset": -102,
@@ -510,6 +525,7 @@ def gusbp3_exon2_end():
         "gene": "GUSBP3",
         "genomic_ac": "NC_000005.10",
         "tx_ac": "NR_027386.2",
+        "strand": -1,
         "seg_start": None,
         "seg_end": {
             "exon_ord": 1,
@@ -534,6 +550,7 @@ def eln_grch38_intronic():
         "gene": "ELN",
         "genomic_ac": "NC_000007.14",
         "tx_ac": "NM_000501.4",
+        "strand": 1,
         "seg_start": {
             "exon_ord": 0,
             "offset": 1,
@@ -569,6 +586,7 @@ def gusbp3_exon5_start():
         "gene": "GUSBP3",
         "genomic_ac": "NC_000005.10",
         "tx_ac": "NR_027386.2",
+        "strand": -1,
         "seg_start": {
             "exon_ord": 4,
             "offset": -3589,
@@ -609,6 +627,7 @@ def genomic_tx_seg_service_checks(actual, expected=None, is_valid=True):
         assert actual.gene == expected.gene
         assert actual.genomic_ac == expected.genomic_ac
         assert actual.tx_ac == expected.tx_ac
+        assert actual.strand == expected.strand
 
         for seg_attr in ["seg_start", "seg_end"]:
             expected_seg = getattr(expected, seg_attr)
@@ -635,6 +654,7 @@ def genomic_tx_seg_service_checks(actual, expected=None, is_valid=True):
         assert actual.gene is None
         assert actual.genomic_ac is None
         assert actual.tx_ac is None
+        assert actual.strand is None
         assert actual.seg_start is None
         assert actual.seg_end is None
         assert len(actual.errors) > 0
@@ -678,6 +698,7 @@ def genomic_tx_seg_checks(actual, expected=None, is_valid=True):
         assert actual.gene == expected.gene
         assert actual.genomic_ac == expected.genomic_ac
         assert actual.tx_ac == expected.tx_ac
+        assert actual.strand == expected.strand
 
         expected_seg = expected.seg
         if expected_seg:
@@ -700,6 +721,7 @@ def genomic_tx_seg_checks(actual, expected=None, is_valid=True):
         assert actual.gene is None
         assert actual.genomic_ac is None
         assert actual.tx_ac is None
+        assert actual.strand is None
         assert actual.seg is None
         assert len(actual.errors) > 0
 
@@ -1463,7 +1485,7 @@ async def test_valid_inputs(test_egc_mapper, eln_grch38_intronic):
 
 
 @pytest.mark.asyncio
-async def test_invalid(test_egc_mapper):
+async def test_invalid(test_egc_mapper, caplog):
     """Test that invalid queries return `None`."""
     resp = await test_egc_mapper.genomic_to_tx_segment(
         transcript="NM_152263 3",
@@ -1507,18 +1529,6 @@ async def test_invalid(test_egc_mapper):
     genomic_tx_seg_service_checks(resp, is_valid=False)
     assert resp.errors == ["Genomic accession does not exist in UTA: NC_000035.200"]
 
-    # Invalid coordinates
-    resp = await test_egc_mapper.genomic_to_tx_segment(
-        genomic_ac="NC_000001.11",
-        seg_start_genomic=9999999999998,
-        seg_end_genomic=9999999999999,
-        transcript="NM_152263.3",
-    )
-    genomic_tx_seg_service_checks(resp, is_valid=False)
-    assert resp.errors == [
-        "9999999999998 on NC_000001.11 does not occur within the exons for NM_152263.3"
-    ]
-
     # Must supply either gene or transcript
     resp = await test_egc_mapper.genomic_to_tx_segment(
         seg_start_genomic=154191901, genomic_ac="NC_000001.11"
@@ -1526,6 +1536,37 @@ async def test_invalid(test_egc_mapper):
     genomic_tx_seg_service_checks(resp, is_valid=False)
     assert resp.errors == ["Must provide either `gene` or `transcript`"]
 
+    # Check 150 bp warning statement
+    with caplog.at_level(logging.WARNING):
+        await test_egc_mapper.genomic_to_tx_segment(
+            genomic_ac="NC_000001.11",
+            seg_start_genomic=9999999999998,
+            seg_end_genomic=9999999999999,
+            transcript="NM_152263.3",
+        )
+    assert any(
+        (
+            "9999999999998 on NC_000001.11 occurs more than 150 bp outside the exon "
+            "boundaries of the NM_152263.3 transcript, indicating this may not be "
+            "a chimeric transcript junction and is unlikely to represent a "
+            "contiguous coding sequence. Confirm that the genomic position 9999999999998 is "
+            "being used to represent transcript junction and not DNA breakpoint."
+        )
+        in record.message
+        for record in caplog.records
+    )
+    assert any(
+        (
+            "9999999999999 on NC_000001.11 occurs more than 150 bp outside the exon "
+            "boundaries of the NM_152263.3 transcript, indicating this may not be "
+            "a chimeric transcript junction and is unlikely to represent a "
+            "contiguous coding sequence. Confirm that the genomic position 9999999999999 is "
+            "being used to represent transcript junction and not DNA breakpoint."
+        )
+        in record.message
+        for record in caplog.records
+    )
+
     # Exon 22 does not exist
     resp = await test_egc_mapper.tx_segment_to_genomic(
         exon_start=None,