Attentive Variational Information Bottleneck for TCR–peptide interaction prediction

• DOI: 10.1093/bioinformatics/btac820
• Publication Date: 2022-12-23T16:34:01Z
• Authors: Filippo Grazioli; Pierre Machart; Anja Mösch; Kai Li; Leonardo V Castorina; Nico Pfeifer; Martin Renqiang Min
• Keywords: Information bottleneck method; Sequence (biology); Code (set theory)

We present a multi-sequence generalization of Variational Information Bottleneck and call the resulting model Attentive (AVIB). Our AVIB leverages multi-head self-attention to implicitly approximate posterior distribution over latent encodings conditioned on multiple input sequences. apply fundamental immuno-oncology problem: predicting interactions between T-cell receptors (TCRs) peptides.Experimental results various datasets show that significantly outperforms state-of-the-art methods for TCR-peptide interaction prediction. Additionally, we learned by is particularly effective unsupervised detection out-of-distribution amino acid sequences.The code data used this study are publicly available at: https://github.com/nec-research/vibtcr.Supplementary at Bioinformatics online.