The dataset is constituted by 204 proteins. The core objective of this dataset is to evaluate the ability of the novel protein descriptors in describing its biomacromolecular structure by predicting the four (All-α,All-β,α/β,α+β) main structural classes.