From a611b4d2f1777b7b32f7a2f0266d36591dd461a1 Mon Sep 17 00:00:00 2001
From: Jeremy Arbesfeld <jarbesfeld@gmail.com>
Date: Thu, 12 Dec 2024 12:12:46 -0500
Subject: [PATCH] Add gene symbol check in translator methods

---
 src/fusor/translator.py   | 18 ++++++++++++++++++
 tests/test_translators.py |  9 +++++++++
 2 files changed, 27 insertions(+)

diff --git a/src/fusor/translator.py b/src/fusor/translator.py
index ca13eb6..cdf825f 100644
--- a/src/fusor/translator.py
+++ b/src/fusor/translator.py
@@ -141,6 +141,21 @@ def _get_gene_element(self, genes: str, caller: Caller) -> GeneElement:
         ge = self.fusor.gene_element(gene=gene)
         return ge[0] if ge[0] else self._get_gene_element_unnormalized(gene)
 
+    def _fusion_symbol_check(self, gene_5prime: str, gene_3prime: str) -> bool:
+        """Check if the normalized gene symbols for the two fusion partners
+        are different. If not, this event is not a fusion
+
+        :param gene_5prime: The 5' gene partner
+        :param gene_3prime: The 3' gene partner
+        :return ``True`` if the gene symbols are different, ``False`` if not
+        """
+        if gene_5prime != gene_3prime:
+            return True
+        _logger.error(
+            "The supplied fusion is not valid as the two fusion partners are the same"
+        )
+        return False
+
     def _get_genomic_ac(self, chrom: str, build: Assembly) -> str:
         """Return a RefSeq genomic accession given a chromosome and a reference build
 
@@ -190,6 +205,9 @@ async def from_jaffa(
         gene_5prime = gene_5prime_element.gene.label
         gene_3prime = gene_3prime_element.gene.label
 
+        if not self._fusion_symbol_check(gene_5prime, gene_3prime):
+            return None
+
         tr_5prime = await self.fusor.transcript_segment_element(
             tx_to_genomic_coords=False,
             genomic_ac=self._get_genomic_ac(chrom1, rb),
diff --git a/tests/test_translators.py b/tests/test_translators.py
index 5eb8c0f..d5199d1 100644
--- a/tests/test_translators.py
+++ b/tests/test_translators.py
@@ -115,6 +115,15 @@ def test_gene_element_arriba(translator_instance):
     assert gene.gene.label == "MIR3672"
 
 
+def test_valid_fusion_partners(translator_instance):
+    """Test that the fusion partners supplied to the translator are different"""
+    partners_check = translator_instance._fusion_symbol_check("BCR", "ABL1")
+    assert partners_check
+
+    partners_check = translator_instance._fusion_symbol_check("BCR", "BCR")
+    assert not partners_check
+
+
 @pytest.mark.asyncio()
 async def test_jaffa(
     fusion_data_example, fusion_data_example_nonexonic, translator_instance