Repertoire sequencing and the statistical ensemble approach to adaptive immunity

The recent advent of high-throughput sequencing of immune receptors allows for the study of immune repertoires in unprecedented depth. This should eventually lead to a better understanding of basic immune function and the development of valuable new diagnostic tools. However, the interpretation of these new sequence data can be difficult because the relationship between receptor sequence and immune specificity is generally unknown. In particular, phenotypically similar repertoires will in general be completely different at the sequence level because of cross-reactivity. Here we argue that sequence repertoires need to be considered statistically to overcome this functional degeneracy. New tools are needed to extract the functionally relevant statistical features from sequence data, separating them from individual-specific, stochastic, and other non-reproducible effects.