Fig. 1: Overview of the analyses used in this study.

RNA is extracted from memory CD4⁺/CD25⁺ regulatory T cells and CD4⁺/CD25^- T cells of type 1 diabetic (T1D) patients and unaffected controls and sequenced. Samples passing QC are aligned to the GRCh37/hg19 human genome (for exonic sequences) and a database of all possible, logical junctions generated from the Aceview (2010 release) human genome annotations. Exons and junctions are quantified in each sample. For each condition (cell type × disease status), exonic sequences and junctions with an average depth per nucleotide (APN) of 2 or greater in at least 50% of samples are considered detected. Detected exons and junctions are summarized to transcripts and transcripts that do not have all their associated exons and junctions detected are filtered out, resulting a reduced set of transcripts per condition. Control- and T1D-specific reduced transcriptomes for each cell type are combined and quantified for each sample of that cell type. For each cell type, transcripts with a transcripts per million (TPM) estimate >0 in 50% of controls and/or T1D cases are considered detected and carried through to data normalization and additional sample QC. Following this, analysis of differential gene expression and differential gene splicing are carried out, and those genes considered statistically significantly different between controls and T1D cases are then further analyzed for functional domain enrichment, gene set enrichment, and additional analyses.