DFWPE: Datasets for Anticancer Peptide Prediction via Dynamic Fourier-Wavelet Positional Encoding

Datasets accompanying the manuscript "DFWPE: Dynamic Fourier-Wavelet Positional Encoding for Anticancer Peptide Prediction via Transfer Learning from General Antimicrobial Peptides" (iScience, under review). Includes: (1) ACP740 benchmark dataset (376 ACP + 364 non-ACP); (2) ACP240 independent test set (120 ACP + 120 non-ACP); (3) AMP pretraining set (5,785 antimicrobial peptides); (4) Negative samples generated from UniProt Swiss-Prot cropping with CD-HIT-2D filtering and k-mer classifier validation (1,500 sequences); (5) Pre-defined 5-fold cross-validation splits for ACP740. Sources: CancerPPD, APD3, TumorHoPe, dbAMP, LAMP2, UniProt Swiss-Prot (release 2024_01).