From 3de425d2321ec188cdfda89b7eb1d8b01cd6f633 Mon Sep 17 00:00:00 2001
From: James Stevenson <james.stevenson@nationwidechildrens.org>
Date: Wed, 3 Jan 2024 16:21:03 -0500
Subject: [PATCH] share method

---
 src/gene/etl/base.py    | 32 +++++++++++++++++++++++++++++++-
 src/gene/etl/ensembl.py | 31 -------------------------------
 src/gene/etl/ncbi.py    | 30 ------------------------------
 3 files changed, 31 insertions(+), 62 deletions(-)

diff --git a/src/gene/etl/base.py b/src/gene/etl/base.py
index 0d8a364e..1dde1585 100644
--- a/src/gene/etl/base.py
+++ b/src/gene/etl/base.py
@@ -7,12 +7,13 @@
 from typing import Dict, List, Optional, Union
 
 import click
+import pandas as pd
 import pydantic
 from biocommons.seqrepo import SeqRepo
 from wags_tails import EnsemblData, HgncData, NcbiGeneData
 
 from gene.database import AbstractDatabase
-from gene.schemas import ITEM_TYPES, Gene, MatchType, SourceName
+from gene.schemas import ITEM_TYPES, Gene, MatchType, SourceName, StoredSequenceLocation
 
 _logger = logging.getLogger(__name__)
 
@@ -206,6 +207,35 @@ def _set_cl_interval_range(self, loc: str, arm_ix: int, location: Dict) -> None:
     #             return chr_location
     #     return None
 
+    def _build_sequence_location(
+        self, seq_id: str, row: pd.Series, concept_id: str
+    ) -> Optional[StoredSequenceLocation]:
+        """Construct a sequence location for storing in a DB.
+
+        :param seq_id: The sequence ID.
+        :param row: A gene from the source file.
+        :param concept_id: record ID from source
+        :return: A storable SequenceLocation containing relevant params for returning a
+        VRS SequenceLocation, or None if unable to retrieve valid parameters
+        """
+        aliases = self._get_seq_id_aliases(seq_id)
+        if not aliases or row.start is None or row.end is None:
+            return None
+
+        sequence = aliases[0]
+
+        if row.start != "." and row.end != "." and sequence:
+            if 0 <= row.start <= row.end:
+                return StoredSequenceLocation(
+                    start=row.start - 1,
+                    end=row.end,
+                    sequence_id=sequence,
+                )
+            else:
+                _logger.warning(
+                    f"{concept_id} has invalid interval: start={row.start - 1} end={row.end}"
+                )
+
     def _get_seq_id_aliases(self, seq_id: str) -> List[str]:
         """Get GA4GH aliases for a sequence id
 
diff --git a/src/gene/etl/ensembl.py b/src/gene/etl/ensembl.py
index 6448d008..201a2145 100644
--- a/src/gene/etl/ensembl.py
+++ b/src/gene/etl/ensembl.py
@@ -14,7 +14,6 @@
     DataLicenseAttributes,
     NamespacePrefix,
     SourceMeta,
-    StoredSequenceLocation,
     Strand,
 )
 
@@ -89,36 +88,6 @@ def _add_gene(self, row: pd.Series, accession_numbers: Dict) -> Dict:
 
         return gene_params
 
-    # TODO reincorporate with NCBI
-    def _build_sequence_location(
-        self, seq_id: str, row: pd.Series, concept_id: str
-    ) -> Optional[StoredSequenceLocation]:
-        """Construct a sequence location for storing in a DB.
-
-        :param seq_id: The sequence ID.
-        :param row: A gene from the source file.
-        :param concept_id: record ID from source
-        :return: A storable SequenceLocation containing relevant params for returning a
-        VRS SequenceLocation, or None if unable to retrieve valid parameters
-        """
-        aliases = self._get_seq_id_aliases(seq_id)
-        if not aliases or row.start is None or row.end is None:
-            return None
-
-        sequence = aliases[0]
-
-        if row.start != "." and row.end != "." and sequence:
-            if 0 <= row.start <= row.end:
-                return StoredSequenceLocation(
-                    start=row.start - 1,
-                    end=row.end,
-                    sequence_id=sequence,
-                )
-            else:
-                _logger.warning(
-                    f"{concept_id} has invalid interval: start={row.start - 1} end={row.end}"
-                )
-
     def _add_attributes(self, row: pd.Series, gene: Dict) -> None:
         """Add concept_id, symbol, and xrefs to a gene record.
 
diff --git a/src/gene/etl/ncbi.py b/src/gene/etl/ncbi.py
index d98681c8..8df2134c 100644
--- a/src/gene/etl/ncbi.py
+++ b/src/gene/etl/ncbi.py
@@ -23,7 +23,6 @@
     NamespacePrefix,
     SourceMeta,
     SourceName,
-    StoredSequenceLocation,
     SymbolStatus,
 )
 
@@ -188,35 +187,6 @@ def _get_gene_info(self, prev_symbols: Dict[str, str]) -> Dict[str, Dict]:
                 info_genes[params["symbol"]] = params
         return info_genes
 
-    def _build_sequence_location(
-        self, seq_id: str, row: pd.Series, concept_id: str
-    ) -> Optional[StoredSequenceLocation]:
-        """Construct a sequence location for storing in a DB.
-
-        :param seq_id: The sequence ID.
-        :param row: A gene from the source file.
-        :param concept_id: record ID from source
-        :return: A storable SequenceLocation containing relevant params for returning a
-        VRS SequenceLocation, or None if unable to retrieve valid parameters
-        """
-        aliases = self._get_seq_id_aliases(seq_id)
-        if not aliases or row.start is None or row.end is None:
-            return None
-
-        sequence = aliases[0]
-
-        if row.start != "." and row.end != "." and sequence:
-            if 0 <= row.start <= row.end:
-                return StoredSequenceLocation(
-                    start=row.start - 1,
-                    end=row.end,
-                    sequence_id=sequence,
-                )
-            else:
-                _logger.warning(
-                    f"{concept_id} has invalid interval: start={row.start - 1} end={row.end}"
-                )
-
     def _get_gene_gff(self, df: pd.DataFrame, info_genes: Dict) -> None:
         """Store genes from NCBI gff file.