Epitope Generation for Peptide-based Cancer Vaccine using Goal-directed Wasserstein Generative Adversarial Network with Gradient Penalty

The folder "generated_peptides" contains the generated peptides from each model tested in the paper in the "Main_paper" folder, and the generated peptides from those models tested in the Supplementary file are in the "Supplementary" folder. The file "Bladder.4.0_test_mut.csv" contains the peptide sequences for bladder cancer used as the training data for the models, and the file "Brain.4.0_test_mut.csv" contains the peptide sequences for bladder cancer. Inside the "Main_paper" folder, each folder corresponds to a model tested in the main text of the paper, and the correspondences of their names to the folder names are listed below. Name of the method in the paper Name of the folder under the Main_paper folder WGAN-GP WGAN-GP MolGAN$^\text{last}$ ($\lambda_M=0.5$) or MolGAN$^\text{best}$ ($\lambda_M=0.5$) MolGAN_lambda_M_0.5 MolGAN$^\text{last}$ ($\lambda_M=0$) or MolGAN$^\text{best}$ ($\lambda_M=0$) MolGAN_lambda_M_0 MolGAN$^\text{last}$ ($\lambda_M=0$) with ORGAN or MolGAN$^\text{best}$ ($\lambda_M=0$) with ORGAN MolGAN_lambda_M_0_ORGAN GD-WGAN-GP ($\gamma_{max}=0.25$) GD-WGAN-GP_gamma_0.25 GD-WGAN-GP ($\gamma_{max}=1$) GD-WGAN-GP_gamma_1 GD-WGAN-GP with ORGAN GD-WGAN-GP_with_ORGAN Each of the above folders has 11 folders with seed0 to 10 appended at the end of the filename. Under each of them, the generated peptides are under the "evaluation_model_last_seed{seed_number}" or "evaluation_model_best_seed{seed_number}" folder, corresponding to the last or best checkpoint (default is last if not mentioned), under the epoch1000 folder. Under those folders, the "generated_raw.csv" contains the 10,000 peptides generated from one of the methods. The peptides after filtering the length to 9 or 10 mers are in "generated_filtered_all.csv". The peptides after length filtering and deduplication are in "generated_filtered_unique.csv". As for the "Supplementary" folder, each folder corresponds to a model tested in the Supplementary file, and the correspondences of their names to the folder names are listed below. Name of the method in the paper Name of the folder under the Supplementary folder Random generator random_generator WGAN-GP WGAN-GP_FixPad_seed0 MolGAN$^\text{last}$ ($\lambda_M=0.5$) or MolGAN$^\text{best}$ ($\lambda_M=0.5$) GD_WGAN_wclip_MolGAN050_DScalar_Loss_MyPredictor_FixPad_seed0 MolGAN$^\text{last}$ ($\lambda_M=0$) or MolGAN$^\text{best}$ ($\lambda_M=0$) GD_WGAN_wclip_MolGAN_Gamma0_DScalar_Loss_MyPredictor_FixPad_seed0 MolGAN$^\text{last}$ ($\lambda_M=0$) with ORGAN or MolGAN$^\text{best}$ ($\lambda_M=0$) with ORGAN GD_WGAN_wclip_OR_PlusOne_MolGAN_Gamma0_DScalar_Loss_MyPredictor_FixPad_seed0 PepINVENT$^\text{best}$ (DeepImmuno direct) or PepINVENT$^\text{last}$ (DeepImmuno direct) PepINVENT_DeepImmuno_direct_seed0 PepINVENT$^\text{best}$ (DeepImmuno with reward network) or PepINVENT$^\text{last}$ (DeepImmuno with reward network) PepINVENT_DeepImmuno_with_scorer_seed0 GD-WGAN-GP ($\gamma_{max}=0.25$) Goal-directed_WGAN-GP_Gamma0_25_FixPad_seed0 GD-WGAN-GP ($\gamma_{max}=0.5$) Goal-directed_WGAN-GP_Gamma0_5_FixPad_seed0 GD-WGAN-GP ($\gamma_{max}=0.75$) Goal-directed_WGAN-GP_Gamma0_75_FixPad_seed0 GD-WGAN-GP ($\gamma_{max}=1$) Goal-directed_WGAN-GP_FixPad_seed0 GD-WGAN-GP with ORGAN Goal-directed_WGAN-GP_ORGAN_PlusOne_FixPad_seed0 GD-WGAN-GP ($\gamma_{max}=1$) with Direct DeepImmuno Scoring Goal-directed_WGAN-GP_noRewardNet_FixPad_seed0 GD-WGAN-GP ($\gamma_{max}=1$ and $S_{scale}=1$) Goal-directed_WGAN-GP_noSscale_FixPad_seed0 GD-WGAN-GP without switch and ORGAN Goal-directed_WGAN-GP_NoSwithORGAN_FixPad_seed0 GD-WGAN-GP with ORGAN and LSTM Goal-directed_WGAN-GP_ORGAN_PlusOne_LSTM_FixPad_seed0 GD-WGAN-GP with ORGAN and Transformer Goal-directed_WGAN-GP_ORGAN_PlusOne_TransformerNoMaskL2_FixPad_seed0 GD-WGAN-GP with switch ($\gamma_{max}=0.25$) and LSTM Goal-directed_WGAN-GP_Switching_Gamma0_25_LSTM_FixPad_seed0 GD-WGAN-GP with switch ($\gamma_{max}=0.25$) and Transformer Goal-directed_WGAN-GP_Switching_Gamma0_25_TransformerNoMaskL2_FixPad_seed0 GD-WGAN-GP with switch ($\gamma_{max}=1$) and LSTM Goal-directed_WGAN-GP_Switching_Gamma1_LSTM_FixPad_seed0 GD-WGAN-GP with switch ($\gamma_{max}=1$) and Transformer Goal-directed_WGAN-GP_Switching_Gamma1_TransformerNoMaskL2_FixPad_seed0 D3PM without reward D3PM_NoReward_AttachedTransformer_seed0 D3PM with ORGAN D3PM_ORGAN_PlusOne_AttachedTransformer_seed0 D3PM with switch ($\gamma_{max}=0.25$) D3PM_ORGAN_Switching_Gamma0_25_AttachedTransformer_seed0 D3PM with switch ($\gamma_{max}=1$) D3PM_ORGAN_Switching_Gamma1_0_AttachedTransformer_seed0