Quantifying selection in high-throughput Immunoglobulin sequencing datasets (58.4)

Abstract

Abstract Quantifying antigen driven selection based on mutation patterns observed in Immunoglobulin (Ig) DNA sequences can provide insights into the basic biology that underlies the adaptive immune response, and may further serve as diagnostic and prognostic markers. High-throughput sequencing approaches make large-scale characterization of B cell Ig repertoires feasible. However, analyzing selection in these large datasets, presents fundamental challenges. Here I will present a new computational framework for Bayesian estimation of Antigen-driven SELectIoN (BASELINe) in Ig sequecnes. BASELINe provides a more intuitive means to analyze selection by shifting the problem from one of detecting selection to one of quantifying it. It allows, for the first time, comparative analysis between sequences derived from different germlines. We have made this method practical for complete repertoire analysis through an online tool (clip.med.yale.edu/baseline) and a highly optimized implementation that can analyze up to 10,000 sequences in under four minutes on a single processor. By testing this method on computer simulations as well as on real sequences from genetically engineered mice, we have shown that indeed there is a significant improvement in our ability to quantify selection strength. We then demonstrate how BASELINe can be applied to high throughput sequencing datasets to arrive at several novel non-trivial observations.