scRNA-seq bioinformatics

Patient samples from the {"type":"entrez-geo","attrs":{"text":"GSE161529","term_id":"161529"}}GSE161529 dataset from Pal and colleagues were downloaded from the Gene Expression Omnibus (GEO) server [7]. We generated a folder containing a barcode, features, and matrix file for each patient sample. We imported seven normal patient sample datasets with the Read10X Seurat command in R. Patient samples N-N280-Epi, N-N1105-Epi, N-MH0064-Epi are premenopausal and nulliparous, N-MH0023-Epi is premenopausal and parous, N-PM0342-Epi is postmenopausal and nulliparous, and N-PM0372-Epi and N-MH275-Epi are postmenopausal and parous [7]. We removed potentially low-quality cells with ≥ 20% mitochondrial genes, < 500 genes, < 1500 unique molecular identifiers (UMI), or < 0.8 log10GenesPerUMI. We removed genes expressed in less than ten cells. SCTransform was used to normalize, scale, and find variable features in the data. Anchor-based integration was performed. We used 100 principal components for PCA analysis, followed by t-SNE and UMAP analysis. SCTransform is highly effective, so 100 principle components contribute to robust analysis. We performed a clustering analysis with several cluster resolution values and generated heatmaps, t-SNE, and dot plots. The top differentially expressed genes were used as population-specific marker genes and examined in detail. We also examined eight scRNAseq TN BrC datasets from {"type":"entrez-geo","attrs":{"text":"GSE161529","term_id":"161529"}}GSE161529 using the above protocols, with the additional step of removing cell populations that highly express CD31 (endothelial cell marker) and CD45 (immune cell marker). Four samples had BRCA mutations: TN-B1-MH4031, TN-B1-MH0131, TN-B1-Tum0554, TN-B1-MH0177; “B1” indicates BRCA mutant. Four samples had normal BRCA: TN-MH0126, TN-MH0135, TN-SH0106, TN-MH0114-T2.