TCR chain type tokens: alpha, gamma, delta and their CDR3 regions

BiomedSciAI · Jun 2, 2024 · b86dc56 · b86dc56
1 parent 14e1b7b
commit b86dc56
Show file tree

Hide file tree

Showing 11 changed files with 16 additions and 16 deletions.
diff --git a/...tokenizer/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json b/...tokenizer/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json
@@ -9008,4 +9008,4 @@
       "c2n c3n("
     ]
   }
-}
+}
diff --git a/...izer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/cell_attributes_tokenizer.json b/...izer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/cell_attributes_tokenizer.json
@@ -3865,4 +3865,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...odulartokenizer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/gene_tokenizer.json b/...odulartokenizer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/gene_tokenizer.json
@@ -97980,4 +97980,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...enizer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/t5_tokenizer_AA_special.json b/...enizer/pretrained_tokenizers/bmfm_extended_modular_tokenizer/t5_tokenizer_AA_special.json
@@ -3102,4 +3102,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...tokenizer/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json b/...tokenizer/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json
@@ -9008,4 +9008,4 @@
       "c2n c3n("
     ]
   }
-}
+}
diff --git a/...ulartokenizer/pretrained_tokenizers/bmfm_modular_tokenizer/cell_attributes_tokenizer.json b/...ulartokenizer/pretrained_tokenizers/bmfm_modular_tokenizer/cell_attributes_tokenizer.json
@@ -3865,4 +3865,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...odulartokenizer/pretrained_tokenizers/bmfm_modular_tokenizer/t5_tokenizer_AA_special.json b/...odulartokenizer/pretrained_tokenizers/bmfm_modular_tokenizer/t5_tokenizer_AA_special.json
@@ -3102,4 +3102,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...ngle_path/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json b/...ngle_path/bpe_tokenizer_trained_on_chembl_zinc_with_aug_4272372_samples_balanced_1_1.json
@@ -9008,4 +9008,4 @@
       "c2n c3n("
     ]
   }
-}
+}
diff --git a/...enizer/pretrained_tokenizers/modular_AA_SMILES_single_path/cell_attributes_tokenizer.json b/...enizer/pretrained_tokenizers/modular_AA_SMILES_single_path/cell_attributes_tokenizer.json
@@ -3865,4 +3865,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/...okenizer/pretrained_tokenizers/modular_AA_SMILES_single_path/t5_tokenizer_AA_special.json b/...okenizer/pretrained_tokenizers/modular_AA_SMILES_single_path/t5_tokenizer_AA_special.json
@@ -3102,4 +3102,4 @@
     },
     "unk_token": "<UNK>"
   }
-}
+}
diff --git a/fusedrug/data/tokenizer/modulartokenizer/special_tokens.py b/fusedrug/data/tokenizer/modulartokenizer/special_tokens.py
@@ -39,12 +39,12 @@
     "MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR3",
     "MOLECULAR_ENTITY_TCR_ALPHA_CHAIN",  # TCR "light" chain - only V, J and C segments (variable region)
     "MOLECULAR_ENTITY_TCR_BETA_VDJ",  # TCR "heavy" chain - V(ariable), D(iversity), and J(oining) segments, as well as the C(onstant) segment
-    "MOLECULAR_ENTITY_TCR_BETA_CDR3", # TCR beta chain CDR3 region
-    "MOLECULAR_ENTITY_TCR_GAMMA_VAR", # TCR gamma chain variable region
-    "MOLECULAR_ENTITY_TCR_DELTA_VAR", # TCR delta chain variable region
-    "MOLECULAR_ENTITY_TCR_ALPHA_CDR3", # TCR alpha chain CDR3 region
-    "MOLECULAR_ENTITY_TCR_GAMMA_CDR3", # TCR gamma chain CDR3 region
-    "MOLECULAR_ENTITY_TCR_DELTA_CDR3", # TCR delta chain CDR3 region
+    "MOLECULAR_ENTITY_TCR_BETA_CDR3",  # TCR beta chain CDR3 region
+    "MOLECULAR_ENTITY_TCR_GAMMA_VAR",  # TCR gamma chain variable region
+    "MOLECULAR_ENTITY_TCR_DELTA_VAR",  # TCR delta chain variable region
+    "MOLECULAR_ENTITY_TCR_ALPHA_CDR3",  # TCR alpha chain CDR3 region
+    "MOLECULAR_ENTITY_TCR_GAMMA_CDR3",  # TCR gamma chain CDR3 region
+    "MOLECULAR_ENTITY_TCR_DELTA_CDR3",  # TCR delta chain CDR3 region
     "TARGETED_ANTIBODY_DESIGN_ENCODER_ONLY_MODE",  # A prefix to our T5 model to inform it that it will run in "encoder only" mode (so only
     # the encoder-stack is used, plus the encoder-output-tokens-classification-head)
     "DECODER_START",
-Original file line number
+Diff line change
@@ Expand Up / @@ -9008,4 +9008,4 @@ @@
           "c2n c3n("
         ]
       }
-    }
+    }