STIG: Generation and simulated sequencing of synthetic T cell receptor repertoires

Abstract T cell receptor repertoire inference from DNA and RNA sequencing experiments is frequently performed to characterize host immune responses to disease states. Existing tools for repertoire inference have been compared across publicly available biological datasets or unpublished simulated sequencing data. Evaluation and comparison of these tools is challenging without common data sets created from a known repertoire with well-defined biological and sequencing characteristics. Here we introduce STIG, a tool to create simulated T cell receptor sequencing data from a customizable virtual T cell repertoire, with clear attribution of individual reads back to locations within their respective T-cell receptor clonotypes. STIG allows for robust performance evaluation of T cell repertoire inference and downstream analysis methods. STIG is implemented in Python 3 and is freely available for download at https://github.com/Benjamin-Vincent-Lab/stig Author summary As part of the acquired immune system, T cells are integral in the host response to microbes, tumors and autoimmune disease. These cells each have a semi-unique T cell receptor that serves to bind a set of antigens that will in turn stimulate that cell to perform its particular pro- (or anti) inflammatory role. This receptor is the product of DNA rearrangement of germline gene segments, similar to B cell receptor loci rearrangement, which provides a wide variety of potential T cell receptors to respond to antigens. At the site of an immune reaction, T cells can increase their number through clonal expansion and methods have been developed to analyze bulk genetic sequencing data to infer the individual receptors and the relative size of their clonal subpopulations present within a sample. To date, these methods and tools have been tested and compared using either biological samples (where the true quantitiy and types of T cells is unknown) or unshared synthetic datasets. In this paper I describe a new tool to generate biologically-inspired T-cell repertoires in-silico and generate simulated sequencing data from them.