SLIDE analysis of the scRNA-seq data using an interpretable factor-analysis machine learning framework, that moves beyond predictive biomarkers to try and infer latent factors underlying LS pathophysiology.

Using transcriptomic profiling at single-cell resolution, we investigate cell-intrinsic and cell-extrinsic signatures associated with pathogenesis and inflammation-driven fibrosis in both adult and pediatric localized Scleroderma (LS) patients. We perform scRNA-seq on adult and pediatric LS patients and healthy controls. We then analyze the scRNA-seq data using an interpretable factor-analysis machine learning framework, SLIDE, that moves beyond predictive biomarkers to try and infer latent factors underlying LS pathophysiology. SLIDE is a novel latent factor regression-based framework that comes with rigorous statistical guarantees regarding identifiability of the latent factors, corresponding inference, and FDR control. We find distinct differences in the characteristics and complexity in the molecular signatures between adult and pediatric LS. SLIDE identified cell type-specific determinants of LS associated with age and severity and revealed insights into signaling mechanisms shared between LS and Systemic Sclerosis (SSc), as well as differences in onset of the disease in the pediatric compared with adult population. Our analyses recapitulate known drivers of LS pathology and find previously unidentified cellular signaling modules that stratify LS subtypes and define a shared signaling axis with SSc.In this study we compared 17 healthy and 27 LS samples evaluated by 10X Genomics single cell sequencing. Samples were processed and clustered for DEG analysis. Overall design: SLIDE analysis of the scRNA-seq data using an interpretable factor-analysis machine learning framework, that moves beyond predictive biomarkers to try and infer latent factors underlying LS pathophysiology. NOTE: The current larger cohort of patients include 44 samples (16 new samples and 28 re-analyzed samples), 27 LS and 17 healthy. These were aggregated using CellRanger and processed via the Seurat pipeline with other R packages including Nichenet, Harmony, etc. The resulting dataset was then analyzed by computational biologists in the Das lab using their SLIDE model. *************************************************************** The table below lists GEO accessions reused/reanalyzed for this study. ***************************************************************