Identification of convergent antibody-specific sequence patterns

摘要

The method uses variational autoencoders (VAEs) and deep generative modeling to learn meaningful representations from immune repertoires. The system can map the input array into a low-dimensional latent space, which reveals a large number of convergent array patterns. The system can identify patterns that reside in converging clusters that are highly predictive of antigen exposure and/or antigen specificity. The system can generate novel functional antibody sequence variants in silico from the latent space. [Selection drawing] Fig. 1