TEST-NEW that was developed in our lab and can be downloaded from here. The dataset comes as a tab-separated text file with three columns that correspond to sequence id (from either TargetDB or PepcDB), annotation of the class label (crystallizable or noncrystallizable), and sequence.