Simulated T cell receptor beta chain sequencing data.

To systematically compare the performance of 12 TCR Rep-Seq analyzing tools including IgBLAST, MiXCR, IMGT/HighVQUEST and so on, we simulated five next-generation sequencing datasets of TCR beta chain. Briefly, we firstly learned from a published deep sequencing dataset (Warren et al. 2011, PMID: 21349924) gene usage, deletion, and insertion distribution, based on which we then built a model for simulation of raw clonotypes. Following that, we employed Zipf’s law to assign size for these clonotypes, and the resulting original clonal repertoire was in-silico amplified for 8 PCR cycles. Finally, we used ART to generate next-generation sequencing data. From the simulated datasets, we can exactly know the V, D, J segments and CDR3 regions of these sequences.