Deconvolution of Donor and Recipient Cells from Lung Transplant Single Cell RNA-seq Data

Abstract

<h3>Purpose</h3> Single-cell RNA-sequencing (scRNA-seq) is a powerful tool to reveal novel insights into transplant biology but a key requirement is the robust identification of donor and recipient cells. Expressed single nucleotide variants (SNVs) can be used to differentiate these cells; however, drastic differences in donor and recipient cell proportions and cell types challenge current methods. We developed "scTx", which mitigates these issues and reliably dentifies donor and recipient cells without matched genotype data. <h3>Methods</h3> scTX, has two steps: 1) identification of donor and recipient genotypes using expressed SNVs, and 2) assignment of each cell to a genotype or cell doublet. We validate our algorithm using simulated lung transplant reperfusion data by mixing scRNA-seq lung data with peripheral blood mononuclear cell data at different ratios and amounts of ambient RNA. We apply our method to four post-transplant bronchoalveolar lavage (BAL) samples and two explanted lungs with chronic lung allograft dysfunction (CLAD). <h3>Results</h3> Using simulated data, scTX can accurately identify donor and recipient cells at low proportions (i.e. < 1% of the total cells) and does not identify a second genotype when one is not present. We show that scTx reliably identifies two genotypes from CLAD and BAL samples and that their donor and recipient origin can be inferred based on annotated cell types (Fig 1). <h3>Conclusion</h3> scTx can reliably deconvolute donor from recipient cells in lung transplant samples with mixed cell types and proportions. This broadly applicable method will unravel mechanisms of rejection and reveal the dynamics of donor and recipient cell populations in human datasets.