cool-seq-tool 0.14.5__tar.gz → 0.15.1__tar.gz

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cool_seq_tool
-Version: 0.14.5
+Version: 0.15.1
 Summary: Common Operation on Lots of Sequences Tool
 Author: Kori Kuzma, James Stevenson, Katie Stahl, Alex Wagner
 License: MIT License
@@ -46,11 +46,11 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: asyncpg
 Requires-Dist: boto3
-Requires-Dist: agct>=0.1.0-dev1
+Requires-Dist: agct>=0.2.0rc1
 Requires-Dist: polars~=1.0
 Requires-Dist: biocommons.seqrepo
 Requires-Dist: pydantic<3.0,>=2.0
-Requires-Dist: ga4gh.vrs<3.0,>=2.1.3
+Requires-Dist: ga4gh.vrs<3.0,>=2.1.4
 Requires-Dist: wags-tails~=0.4.0
 Requires-Dist: bioutils
 Provides-Extra: dev

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/pyproject.toml

@@ -26,11 +26,11 @@ license = {file = "LICENSE"}
 dependencies = [
     "asyncpg",
     "boto3",
-    "agct >= 0.1.0-dev1",
+    "agct >= 0.2.0rc1",
     "polars ~= 1.0",
     "biocommons.seqrepo",
     "pydantic >=2.0,<3.0",
-    "ga4gh.vrs >=2.1.3,<3.0",
+    "ga4gh.vrs >=2.1.4,<3.0",
     "wags-tails ~= 0.4.0",
     "bioutils",
 ]

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool/app.py

@@ -107,6 +107,7 @@ class CoolSeqTool:
         self.ex_g_coords_mapper = ExonGenomicCoordsMapper(
             self.seqrepo_access,
             self.uta_db,
+            self.mane_transcript,
             self.mane_transcript_mappings,
             self.liftover,
         )

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool/mappers/exon_genomic_coords.py

@@ -8,12 +8,15 @@ from pydantic import ConfigDict, Field, StrictInt, StrictStr, model_validator
 
 from cool_seq_tool.handlers.seqrepo_access import SeqRepoAccess
 from cool_seq_tool.mappers.liftover import LiftOver
+from cool_seq_tool.mappers.mane_transcript import ManeTranscript
 from cool_seq_tool.schemas import (
+    AnnotationLayer,
     Assembly,
     BaseModelForbidExtra,
     CoordinateType,
     ServiceMeta,
     Strand,
+    TranscriptPriority,
 )
 from cool_seq_tool.sources.mane_transcript_mappings import ManeTranscriptMappings
 from cool_seq_tool.sources.uta_database import GenomicAlnData, UtaDatabase
@@ -113,6 +116,9 @@ class GenomicTxSeg(BaseModelForbidExtra):
     )
     genomic_ac: StrictStr | None = Field(None, description="RefSeq genomic accession.")
     tx_ac: StrictStr | None = Field(None, description="RefSeq transcript accession.")
+    tx_status: TranscriptPriority | None = Field(
+        None, description="Transcript priority for RefSeq transcript accession"
+    )
     strand: Strand | None = Field(
         None, description="The strand that the transcript accession exists on."
     )
@@ -144,6 +150,7 @@ class GenomicTxSeg(BaseModelForbidExtra):
                 "gene": "TPM3",
                 "genomic_ac": "NC_000001.11",
                 "tx_ac": "NM_152263.3",
+                "tx_status": "longest_compatible_remaining",
                 "strand": -1,
                 "seg": {
                     "exon_ord": 0,
@@ -172,6 +179,9 @@ class GenomicTxSegService(BaseModelForbidExtra):
     )
     genomic_ac: StrictStr | None = Field(None, description="RefSeq genomic accession.")
     tx_ac: StrictStr | None = Field(None, description="RefSeq transcript accession.")
+    tx_status: TranscriptPriority | None = Field(
+        None, description="Transcript priority for RefSeq transcript accession"
+    )
     strand: Strand | None = Field(
         None, description="The strand that the transcript exists on."
     )
@@ -211,6 +221,7 @@ class GenomicTxSegService(BaseModelForbidExtra):
                 "gene": "TPM3",
                 "genomic_ac": "NC_000001.11",
                 "tx_ac": "NM_152263.3",
+                "tx_status": "longest_compatible_remaining",
                 "strand": -1,
                 "seg_start": {
                     "exon_ord": 0,
@@ -264,6 +275,7 @@ class ExonGenomicCoordsMapper:
         self,
         seqrepo_access: SeqRepoAccess,
         uta_db: UtaDatabase,
+        mane_transcript: ManeTranscript,
         mane_transcript_mappings: ManeTranscriptMappings,
         liftover: LiftOver,
     ) -> None:
@@ -288,11 +300,13 @@ class ExonGenomicCoordsMapper:
 
         :param seqrepo_access: SeqRepo instance to give access to query SeqRepo database
         :param uta_db: UtaDatabase instance to give access to query UTA database
+        :param mane_transcript: ManeTranscript instance to give access to ManeTranscript class
         :param mane_transcript_mappings: Instance to provide access to ManeTranscriptMappings class
         :param liftover: Instance to provide mapping between human genome assemblies
         """
         self.seqrepo_access = seqrepo_access
         self.uta_db = uta_db
+        self.mane_transcript = mane_transcript
         self.mane_transcript_mappings = mane_transcript_mappings
         self.liftover = liftover
 
@@ -431,6 +445,7 @@ class ExonGenomicCoordsMapper:
             gene=gene,
             genomic_ac=genomic_ac,
             tx_ac=transcript,
+            tx_status=self.mane_transcript_mappings.get_transcript_status(transcript),
             strand=strand,
             seg_start=seg_start,
             seg_end=seg_end,
@@ -522,6 +537,7 @@ 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["tx_status"] = start_tx_seg_data.tx_status
             params["strand"] = start_tx_seg_data.strand
             params["seg_start"] = start_tx_seg_data.seg
         else:
@@ -557,6 +573,7 @@ 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["tx_status"] = end_tx_seg_data.tx_status
                 params["strand"] = end_tx_seg_data.strand
 
             params["seg_end"] = end_tx_seg_data.seg
@@ -858,14 +875,18 @@ class ExonGenomicCoordsMapper:
             if mane_transcripts:
                 transcript = mane_transcripts[0]["RefSeq_nuc"]
             else:
-                # Attempt to find a coding transcript if a MANE transcript
+                # Attempt to find longest compatible transcript if a MANE transcript
                 # cannot be found
-                results = await self.uta_db.get_transcripts(
-                    gene=gene, alt_ac=genomic_ac
+                results = await self.mane_transcript.get_longest_compatible_transcript(
+                    start_pos=genomic_pos,
+                    end_pos=genomic_pos,
+                    gene=gene,
+                    alt_ac=genomic_ac,
+                    start_annotation_layer=AnnotationLayer.GENOMIC,
                 )
 
-                if not results.is_empty():
-                    transcript = results[0]["tx_ac"][0]
+                if results:
+                    transcript = results.refseq
                 else:
                     # Run if gene is for a noncoding transcript
                     query = f"""
@@ -962,6 +983,7 @@ class ExonGenomicCoordsMapper:
             gene=gene,
             genomic_ac=genomic_ac,
             tx_ac=transcript,
+            tx_status=self.mane_transcript_mappings.get_transcript_status(transcript),
             strand=strand,
             seg=TxSegment(
                 exon_ord=exon_num,

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool/mappers/liftover.py

@@ -6,7 +6,8 @@ Currently only supports GRCh37 <-> GRCh38
 import logging
 from os import environ
 
-from agct import Converter, Genome
+from agct import Assembly as AgctAssembly
+from agct import Converter
 
 from cool_seq_tool.schemas import Assembly
 from cool_seq_tool.utils import process_chromosome_input
@@ -43,13 +44,13 @@ class LiftOver:
         """
         self.from_37_to_38 = Converter(
             chainfile=chain_file_37_to_38 or LIFTOVER_CHAIN_37_TO_38,
-            from_db=Genome.HG19,
-            to_db=Genome.HG38,
+            from_assembly=AgctAssembly.HG19,
+            to_assembly=AgctAssembly.HG38,
         )
         self.from_38_to_37 = Converter(
             chainfile=chain_file_38_to_37 or LIFTOVER_CHAIN_38_TO_37,
-            from_db=Genome.HG38,
-            to_db=Genome.HG19,
+            from_assembly=AgctAssembly.HG38,
+            to_assembly=AgctAssembly.HG19,
         )
 
     def get_liftover(
@@ -77,9 +78,9 @@ class LiftOver:
         """
         chromosome = process_chromosome_input(chromosome, "LiftOver.get_liftover()")
         if liftover_to_assembly == Assembly.GRCH38:
-            liftover = self.from_37_to_38.convert_coordinate(chromosome, pos)
+            liftover = self.from_37_to_38.convert_coordinate(chromosome, pos, pos)
         elif liftover_to_assembly == Assembly.GRCH37:
-            liftover = self.from_38_to_37.convert_coordinate(chromosome, pos)
+            liftover = self.from_38_to_37.convert_coordinate(chromosome, pos, pos)
         else:
             _logger.warning("%s assembly not supported", liftover_to_assembly)
             liftover = None

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool/resources/status.py

@@ -114,9 +114,9 @@ async def check_status(
             chain_file_38_to_37=chain_file_38_to_37,
         )
     except (FileNotFoundError, ChainfileError):
-        _logger.exception("agct converter setup failed")
+        _logger.exception("`agct` converter setup failed")
     except Exception as e:
-        _logger.critical("Encountered unexpected error setting up agct: %s", e)
+        _logger.critical("Encountered unexpected error setting up `agct`: %s", e)
     else:
         status["liftover"] = True
 

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool/sources/mane_transcript_mappings.py

@@ -8,7 +8,7 @@ from pathlib import Path
 import polars as pl
 
 from cool_seq_tool.resources.data_files import DataFile, get_data_file
-from cool_seq_tool.schemas import ManeGeneData
+from cool_seq_tool.schemas import ManeGeneData, TranscriptPriority
 
 _logger = logging.getLogger(__name__)
 
@@ -85,6 +85,22 @@ class ManeTranscriptMappings:
             return []
         return mane_rows.to_dicts()
 
+    def get_transcript_status(self, tx_ac: str) -> TranscriptPriority:
+        """Get MANE status for a transcript
+
+        :param tx_ac: A RefSeq transcript accession
+        :return: A TranscriptPriority object
+        """
+        mane_info = self.get_mane_from_transcripts([tx_ac])
+        if not mane_info:
+            return TranscriptPriority.LONGEST_COMPATIBLE_REMAINING
+        mane_info = mane_info[0]["MANE_status"]
+        return (
+            TranscriptPriority.MANE_SELECT
+            if mane_info == "MANE Select"
+            else TranscriptPriority.MANE_PLUS_CLINICAL
+        )
+
     def get_mane_data_from_chr_pos(
         self, alt_ac: str, start: int, end: int
     ) -> list[dict]:

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cool_seq_tool
-Version: 0.14.5
+Version: 0.15.1
 Summary: Common Operation on Lots of Sequences Tool
 Author: Kori Kuzma, James Stevenson, Katie Stahl, Alex Wagner
 License: MIT License
@@ -46,11 +46,11 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: asyncpg
 Requires-Dist: boto3
-Requires-Dist: agct>=0.1.0-dev1
+Requires-Dist: agct>=0.2.0rc1
 Requires-Dist: polars~=1.0
 Requires-Dist: biocommons.seqrepo
 Requires-Dist: pydantic<3.0,>=2.0
-Requires-Dist: ga4gh.vrs<3.0,>=2.1.3
+Requires-Dist: ga4gh.vrs<3.0,>=2.1.4
 Requires-Dist: wags-tails~=0.4.0
 Requires-Dist: bioutils
 Provides-Extra: dev

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/src/cool_seq_tool.egg-info/requires.txt

@@ -1,10 +1,10 @@
 asyncpg
 boto3
-agct>=0.1.0-dev1
+agct>=0.2.0rc1
 polars~=1.0
 biocommons.seqrepo
 pydantic<3.0,>=2.0
-ga4gh.vrs<3.0,>=2.1.3
+ga4gh.vrs<3.0,>=2.1.4
 wags-tails~=0.4.0
 bioutils
 

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/tests/mappers/test_exon_genomic_coords.py

@@ -172,6 +172,7 @@ def tpm3_exon1():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "tx_status": "longest_compatible_remaining",
         "strand": -1,
         "seg": {
             "exon_ord": 0,
@@ -197,6 +198,7 @@ def tpm3_exon8():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "tx_status": "longest_compatible_remaining",
         "strand": -1,
         "seg": {
             "exon_ord": 7,
@@ -222,6 +224,7 @@ def tpm3_exon1_g(tpm3_exon1):
         "gene": tpm3_exon1.gene,
         "genomic_ac": tpm3_exon1.genomic_ac,
         "tx_ac": tpm3_exon1.tx_ac,
+        "tx_status": tpm3_exon1.tx_status,
         "strand": tpm3_exon1.strand,
         "seg_start": tpm3_exon1.seg,
     }
@@ -235,6 +238,7 @@ def tpm3_exon8_g(tpm3_exon8):
         "gene": tpm3_exon8.gene,
         "genomic_ac": tpm3_exon8.genomic_ac,
         "tx_ac": tpm3_exon8.tx_ac,
+        "tx_status": tpm3_exon8.tx_status,
         "strand": tpm3_exon8.strand,
         "seg_end": tpm3_exon8.seg,
     }
@@ -248,6 +252,7 @@ def tpm3_exon1_exon8(tpm3_exon1, tpm3_exon8):
         "gene": tpm3_exon8.gene,
         "genomic_ac": tpm3_exon8.genomic_ac,
         "tx_ac": tpm3_exon8.tx_ac,
+        "tx_status": tpm3_exon8.tx_status,
         "strand": tpm3_exon8.strand,
         "seg_start": tpm3_exon1.seg,
         "seg_end": tpm3_exon8.seg,
@@ -269,6 +274,7 @@ def tpm3_exon1_exon8_offset(tpm3_exon1, tpm3_exon8):
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.3",
+        "tx_status": "longest_compatible_remaining",
         "strand": -1,
         "seg_start": tpm3_exon1_cpy.seg,
         "seg_end": tpm3_exon8_cpy.seg,
@@ -283,6 +289,7 @@ def mane_braf():
         "gene": "BRAF",
         "genomic_ac": "NC_000007.14",
         "tx_ac": "NM_004333.6",
+        "tx_status": "mane_select",
         "strand": -1,
         "seg_start": {
             "exon_ord": 5,
@@ -321,6 +328,7 @@ def wee1_exon2_exon11():
         "gene": "WEE1",
         "genomic_ac": "NC_000011.10",
         "tx_ac": "NM_003390.3",
+        "tx_status": "longest_compatible_remaining",
         "strand": 1,
         "seg_start": {
             "exon_ord": 1,
@@ -359,6 +367,7 @@ def mane_wee1_exon2_exon11():
         "gene": "WEE1",
         "genomic_ac": "NC_000011.10",
         "tx_ac": "NM_003390.4",
+        "tx_status": "mane_select",
         "strand": 1,
         "seg_start": {
             "exon_ord": 1,
@@ -397,6 +406,7 @@ def ntrk1_exon10_exon17():
         "gene": "NTRK1",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_002529.3",
+        "tx_status": "longest_compatible_remaining",
         "strand": 1,
         "seg_start": {
             "exon_ord": 9,
@@ -435,6 +445,7 @@ def zbtb10_exon3_end():
         "gene": "ZBTB10",
         "genomic_ac": "NC_000008.11",
         "tx_ac": "NM_001105539.3",
+        "tx_status": "mane_select",
         "strand": 1,
         "seg_start": None,
         "seg_end": {
@@ -461,6 +472,7 @@ def zbtb10_exon5_start():
         "gene": "ZBTB10",
         "genomic_ac": "NC_000008.11",
         "tx_ac": "NM_001105539.3",
+        "tx_status": "mane_select",
         "strand": 1,
         "seg_start": {
             "exon_ord": 4,
@@ -487,6 +499,7 @@ def tpm3_exon6_end():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.4",
+        "tx_status": "mane_select",
         "strand": -1,
         "seg_start": None,
         "seg_end": {
@@ -513,6 +526,7 @@ def tpm3_exon5_start():
         "gene": "TPM3",
         "genomic_ac": "NC_000001.11",
         "tx_ac": "NM_152263.4",
+        "tx_status": "mane_select",
         "strand": -1,
         "seg_start": {
             "exon_ord": 4,
@@ -539,6 +553,7 @@ def gusbp3_exon2_end():
         "gene": "GUSBP3",
         "genomic_ac": "NC_000005.10",
         "tx_ac": "NR_027386.2",
+        "tx_status": "longest_compatible_remaining",
         "strand": -1,
         "seg_start": None,
         "seg_end": {
@@ -565,6 +580,7 @@ def eln_grch38_intronic():
         "gene": "ELN",
         "genomic_ac": "NC_000007.14",
         "tx_ac": "NM_000501.4",
+        "tx_status": "mane_select",
         "strand": 1,
         "seg_start": {
             "exon_ord": 0,
@@ -603,6 +619,7 @@ def gusbp3_exon5_start():
         "gene": "GUSBP3",
         "genomic_ac": "NC_000005.10",
         "tx_ac": "NR_027386.2",
+        "tx_status": "longest_compatible_remaining",
         "strand": -1,
         "seg_start": {
             "exon_ord": 4,
@@ -645,6 +662,7 @@ 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.tx_status == expected.tx_status
         assert actual.strand == expected.strand
 
         for seg_attr in ["seg_start", "seg_end"]:
@@ -676,6 +694,7 @@ 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.tx_status is None
         assert actual.strand is None
         assert actual.seg_start is None
         assert actual.seg_end is None
@@ -720,6 +739,7 @@ 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.tx_status == expected.tx_status
         assert actual.strand == expected.strand
 
         expected_seg = expected.seg
@@ -747,6 +767,7 @@ 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.tx_status is None
         assert actual.strand is None
         assert actual.seg is None
         assert len(actual.errors) > 0
@@ -1050,6 +1071,17 @@ async def test_genomic_to_transcript_fusion_context(
     resp = await test_egc_mapper.genomic_to_tx_segment(**inputs)
     genomic_tx_seg_service_checks(resp, gusbp3_exon5_start)
 
+    # Test case where gene does not have a MANE transcript. We are looking
+    # to check that the same transcript accession is returned across runs
+    inputs = {
+        "genomic_ac": "NC_000001.11",
+        "seg_end_genomic": 156421555,
+        "gene": "MIR9-1HG",
+    }
+    resp = await test_egc_mapper.genomic_to_tx_segment(**inputs)
+    assert resp.tx_ac == "NM_001320454.2"
+    assert resp.tx_status == "longest_compatible_remaining"
+
 
 @pytest.mark.asyncio
 async def test_get_alt_ac_start_and_end(

{cool_seq_tool-0.14.5 → cool_seq_tool-0.15.1}/tests/sources/test_mane_transcript_mappings.py

@@ -5,7 +5,7 @@ from unittest.mock import patch
 import polars as pl
 import pytest
 
-from cool_seq_tool.schemas import ManeGeneData
+from cool_seq_tool.schemas import ManeGeneData, TranscriptPriority
 
 
 @pytest.fixture(scope="module")
@@ -168,6 +168,15 @@ def test_get_mane_from_transcripts(
     assert resp == []
 
 
+def test_get_transcript_status(test_mane_transcript_mappings):
+    """Test that get_transcript_status works correctly"""
+    actual = test_mane_transcript_mappings.get_transcript_status("NM_152263.4")
+    assert actual == TranscriptPriority.MANE_SELECT
+
+    actual = test_mane_transcript_mappings.get_transcript_status("NM_152263.3")
+    assert actual == TranscriptPriority.LONGEST_COMPATIBLE_REMAINING
+
+
 def test_get_mane_data_from_chr_pos(
     test_mane_transcript_mappings, braf_select, braf_plus_clinical
 ):