Statistical inference of the generation probability of T-cell receptors from sequence repertoires

摘要

Stochastic rearrangement of germline DNA by VDJ recombination is at theorigin of immune system diversity. This process is implemented via a series ofstochastic molecular events involving gene choices and random nucleotideinsertions between, and deletions from, genes. We use large sequencerepertoires of the variable CDR3 region of human CD4+ T-cell receptor betachains to infer the statistical properties of these basic biochemical events.Since any given CDR3 sequence can be produced in multiple ways, the probabilitydistribution of hidden recombination events cannot be inferred directly fromthe observed sequences; we therefore develop a maximum likelihood inferencemethod to achieve this end. To separate the properties of the molecularrearrangement mechanism from the effects of selection, we focus onnon-productive CDR3 sequences in T-cell DNA. We infer the joint distribution ofthe various generative events that occur when a new T-cell receptor gene iscreated. We find a rich picture of correlation (and absence thereof), providinginsight into the molecular mechanisms involved. The generative event statisticsare consistent between individuals, suggesting a universal biochemical process.Our distribution predicts the generation probability of any specific CDR3sequence by the primitive recombination process, allowing us to quantify thepotential diversity of the T-cell repertoire and to understand why somesequences are shared between individuals. We argue that the use of formalstatistical inference methods, of the kind presented in this paper, will beessential for quantitative understanding of the generation and evolution ofdiversity in the adaptive immune system.