Datasets used to train/test BetAware-Deep

Training dataset

This dataset contains overall 127 proteins, including 58 TMBB proteins and 69 non-TM beta-barrel proteins (used as negative set). These data have been used to train/test BetAware-Deep in cross-validation. We provide all annotations as well as cross-validation splits.

Independent test dataset

This dataset comprises 15 TMBB proteins used for independent testing and comparison with other approaches.

Discrimination dataset

This dataset comprises 8580 proteins (1009 TMBB and 7571 non-TMBB). The dataset has been originally presented in the following publication:

Tsirigos et al. (2016) PRED-TMBB2: improved topology prediction and detection of beta-barrel outer membrane proteins. Bioinformatics, 32(17), i665-i671.