Datasets used for training, validating and testing "TCRCluster: A Novel Approach to T-Cell Receptor latent featurization and Clustering Using Contrastive Learning-Guided Two-Stage Variational Autoencoders"

The binders_only dataset contain the positive only dataset that have been used in training the one-stage models as well as for clustering The "swapped_negs" dataset contains swapped negatives and have been used in training the two-stage models as well as doing the TCRBase experiment The "train all partitions" dataset contains the the 5 partitions, where partition 2,3,4 were used for training, partition 0 was used for validation and partition 1 was used as held-out external testing. The "17 peps" subsets were used in the TCRbase and clustering experiments.