diff --git a/Makefile b/Makefile index 32ca113..ccdb605 100644 --- a/Makefile +++ b/Makefile @@ -7,7 +7,7 @@ PYTHON_VERSION = 3.8 PYTHON_INTERPRETER = python # --- CLI flag 变量 --- -MPNN_FLAG = $(if $(USE_MPNN),--use-mpnn,) +MPNN_FLAG = $(if $(NO_MPNN),,--use-mpnn) DEVICE_FLAG = $(if $(DEVICE),--device $(DEVICE),) SEED_FLAG = $(if $(SEED),--seed $(SEED),) N_TRIALS_FLAG = $(if $(N_TRIALS),--n-trials $(N_TRIALS),) diff --git a/data/processed/benchmark/feature_columns.txt b/data/processed/benchmark/feature_columns.txt new file mode 100644 index 0000000..09d6735 --- /dev/null +++ b/data/processed/benchmark/feature_columns.txt @@ -0,0 +1,55 @@ +smiles +Cationic_Lipid_to_mRNA_weight_ratio +Cationic_Lipid_Mol_Ratio +Phospholipid_Mol_Ratio +Cholesterol_Mol_Ratio +PEG_Lipid_Mol_Ratio +Purity_Pure +Purity_Crude +Mix_type_Microfluidic +Mix_type_Pipetting +Cargo_type_mRNA +Cargo_type_pDNA +Cargo_type_siRNA +Target_or_delivered_gene_FFL +Target_or_delivered_gene_Peptide_barcode +Target_or_delivered_gene_hEPO +Target_or_delivered_gene_FVII +Target_or_delivered_gene_GFP +Helper_lipid_ID_DOPE +Helper_lipid_ID_DOTAP +Helper_lipid_ID_DSPC +Helper_lipid_ID_MDOA +Model_type_A549 +Model_type_BDMC +Model_type_BMDM +Model_type_HBEC_ALI +Model_type_HEK293T +Model_type_HeLa +Model_type_IGROV1 +Model_type_Mouse +Model_type_RAW264p7 +Delivery_target_body +Delivery_target_dendritic_cell +Delivery_target_generic_cell +Delivery_target_liver +Delivery_target_lung +Delivery_target_lung_epithelium +Delivery_target_macrophage +Delivery_target_muscle +Delivery_target_spleen +Route_of_administration_in_vitro +Route_of_administration_intramuscular +Route_of_administration_intratracheal +Route_of_administration_intravenous +Batch_or_individual_or_barcoded_Barcoded +Batch_or_individual_or_barcoded_Individual +Value_name_log_luminescence +Value_name_luminescence +Value_name_FFL_silencing +Value_name_Peptide_abundance +Value_name_hEPO +Value_name_FVII_silencing +Value_name_GFP_delivery +Value_name_Discretized_luminescence +quantified_delivery \ No newline at end of file diff --git a/data/processed/benchmark/fold_0/test.parquet b/data/processed/benchmark/fold_0/test.parquet new file mode 100644 index 0000000..064ece0 Binary files /dev/null and b/data/processed/benchmark/fold_0/test.parquet differ diff --git a/data/processed/benchmark/fold_0/train.parquet b/data/processed/benchmark/fold_0/train.parquet new file mode 100644 index 0000000..f593aea Binary files /dev/null and b/data/processed/benchmark/fold_0/train.parquet differ diff --git a/data/processed/benchmark/fold_0/valid.parquet b/data/processed/benchmark/fold_0/valid.parquet new file mode 100644 index 0000000..dbf73d9 Binary files /dev/null and b/data/processed/benchmark/fold_0/valid.parquet differ diff --git a/data/processed/benchmark/fold_1/test.parquet b/data/processed/benchmark/fold_1/test.parquet new file mode 100644 index 0000000..dbf73d9 Binary files /dev/null and b/data/processed/benchmark/fold_1/test.parquet differ diff --git a/data/processed/benchmark/fold_1/train.parquet b/data/processed/benchmark/fold_1/train.parquet new file mode 100644 index 0000000..3c1b676 Binary files /dev/null and b/data/processed/benchmark/fold_1/train.parquet differ diff --git a/data/processed/benchmark/fold_1/valid.parquet b/data/processed/benchmark/fold_1/valid.parquet new file mode 100644 index 0000000..5572ce3 Binary files /dev/null and b/data/processed/benchmark/fold_1/valid.parquet differ diff --git a/data/processed/benchmark/fold_2/test.parquet b/data/processed/benchmark/fold_2/test.parquet new file mode 100644 index 0000000..5572ce3 Binary files /dev/null and b/data/processed/benchmark/fold_2/test.parquet differ diff --git a/data/processed/benchmark/fold_2/train.parquet b/data/processed/benchmark/fold_2/train.parquet new file mode 100644 index 0000000..3679fb2 Binary files /dev/null and b/data/processed/benchmark/fold_2/train.parquet differ diff --git a/data/processed/benchmark/fold_2/valid.parquet b/data/processed/benchmark/fold_2/valid.parquet new file mode 100644 index 0000000..f91efae Binary files /dev/null and b/data/processed/benchmark/fold_2/valid.parquet differ diff --git a/data/processed/benchmark/fold_3/test.parquet b/data/processed/benchmark/fold_3/test.parquet new file mode 100644 index 0000000..f91efae Binary files /dev/null and b/data/processed/benchmark/fold_3/test.parquet differ diff --git a/data/processed/benchmark/fold_3/train.parquet b/data/processed/benchmark/fold_3/train.parquet new file mode 100644 index 0000000..f2834b8 Binary files /dev/null and b/data/processed/benchmark/fold_3/train.parquet differ diff --git a/data/processed/benchmark/fold_3/valid.parquet b/data/processed/benchmark/fold_3/valid.parquet new file mode 100644 index 0000000..d41cc4b Binary files /dev/null and b/data/processed/benchmark/fold_3/valid.parquet differ diff --git a/data/processed/benchmark/fold_4/test.parquet b/data/processed/benchmark/fold_4/test.parquet new file mode 100644 index 0000000..d41cc4b Binary files /dev/null and b/data/processed/benchmark/fold_4/test.parquet differ diff --git a/data/processed/benchmark/fold_4/train.parquet b/data/processed/benchmark/fold_4/train.parquet new file mode 100644 index 0000000..f214720 Binary files /dev/null and b/data/processed/benchmark/fold_4/train.parquet differ diff --git a/data/processed/benchmark/fold_4/valid.parquet b/data/processed/benchmark/fold_4/valid.parquet new file mode 100644 index 0000000..064ece0 Binary files /dev/null and b/data/processed/benchmark/fold_4/valid.parquet differ diff --git a/data/processed/train_pretrain.parquet b/data/processed/train_pretrain.parquet index 3bd3047..779ee59 100644 Binary files a/data/processed/train_pretrain.parquet and b/data/processed/train_pretrain.parquet differ diff --git a/data/processed/val_pretrain.parquet b/data/processed/val_pretrain.parquet index f9a4cc0..c478e9d 100644 Binary files a/data/processed/val_pretrain.parquet and b/data/processed/val_pretrain.parquet differ diff --git a/models/benchmark/config.json b/models/benchmark/config.json new file mode 100644 index 0000000..9405c1c --- /dev/null +++ b/models/benchmark/config.json @@ -0,0 +1,15 @@ +{ + "d_model": 256, + "num_heads": 8, + "n_attn_layers": 4, + "fusion_strategy": "attention", + "head_hidden_dim": 128, + "dropout": 0.1, + "use_mpnn": false, + "mpnn_ensemble_paths": null, + "lr": 0.0001, + "weight_decay": 1e-05, + "batch_size": 64, + "epochs": 50, + "patience": 10 +} \ No newline at end of file