Inhibition of epithelial cell YAP-TEAD/LOX signaling attenuates pulmonary fibrosis "

SnRNAseq analysis of PCLS treated with Verteporfin in fibrotic conditions

After preprocessing, analysis of the ex vivo human PCLS snRNA-seq data was conducted using the Seurat package (version 1.8.2). Cells with less than 1000 transcripts profiled were then removed.

To minimize the possible effect of potential batch correction methods, we first processed and annotated each library  separately, before integrating them together and annotating them jointly. To integrate the multiple snRNA-seq datasets, we employed Robust Principal Component Analysis (RPCA). RPCA is a powerful technique for decomposing a data matrix into low-rank and sparse components. Briefly, the low-rank component represents shared biological signals across datasets, while the sparse component captures dataset-specific variations and technical noise. Based on the cellular diversity, we chose to use PCLS treated with DMSO as the reference for the integration. Following the RPCA decomposition, we utilized the low-rank component as the integrated representation of the snRNA-seq datasets. This component captured shared biological signals across conditions while mitigating dataset-specific variations.However,  subsequent analyses, such as clustering and differential expression analysis, were performed on the non integrated  but normalized gene expression values. To validate the effectiveness of the integrated representation, we performed various analyses, including cell-type clustering, identification of marker genes. We also compared the results of these analyses to those obtained from individual datasets to evaluate the improvement gained through the integration process. Marker genes were computed using a Wilcoxon rank-sum test, and genes were considered marker genes if the FDR-corrected p-value was below 0.05 and the log2 fold change was above 0.5.