An immune-focused supplemental alignment pipeline captures information missed from dominant single-cell RNA-seq analyses, including allele-specific MHC-I regulation

Sebastian Benjamin¹GW Mcelfresh¹Maanasa Kaza¹Gregory J Boggy¹Benjamin Varco-Merth¹Sohita Ojha¹Shana Feltham¹William Goodwin¹Candice Nkoy¹Derick Duell¹Andrea Selseth¹Tyler Bennett¹Aaron Barber-Axthelm¹Nicole N Haese¹Helen Wu¹Courtney Waytashek¹Carla Boyle¹Jeremy Smedley¹Caralyn S Labriola¹Michael Axthelm¹R. Keith Reeves²Daniel N Streblow¹Jonah B Sacha¹Afam Okoye¹Scott G Hansen¹Louis J Picker¹Benjamin Bimber^1*

• ¹Oregon Health and Science University, Portland, United States
• ²School of Medicine, Duke University, Durham, North Carolina, United States

RNA sequencing (RNA-seq) can measure whole transcriptome gene expression from tissues or even individual cells, providing a powerful tool to study the immune response. Analysis of RNA-seq data involves mapping relatively short sequence reads to a reference genome, and quantifying genes based on the position of alignments relative to annotated genes. While this is usually robust, genetic polymorphism or genome/annotation inaccuracies result in genes with systematically missing or inaccurate data. These issues are frequently hidden or ignored, yet are highly relevant to immunologic data, where balancing selection has generated many polygenic gene families not accurately represented in a 'one-size-fits-all' reference genome.Here we present nimble, a tool to supplement standard RNA-seq pipelines. Nimble uses a previously developed pseudoaligner to process either bulk-or single-cell RNA-seq data using custom gene spaces.Importantly, nimble can apply customizable scoring criteria to each gene set, tailored to the biology of those genes. We demonstrate that nimble recovers data in diverse contexts, ranging from simple cases (e.g., incorrect gene annotation or viral RNA), to complex immune genotyping (e.g., major histocompatibility or killerimmunoglobulin-like receptors). We use this enhanced capability to identify killer-immunoglobulin-like receptor expression specific to tissue-resident memory T cells and demonstrate allele-specific regulation of MHC alleles after Mycobacterium tuberculosis stimulation. Combining nimble data with standard pipelines enhances the fidelity and accuracy of experiments, maximizing the value of expensive datasets, and identifying cellular subsets not possible with standard tools alone.