Single-cell RNA-seq analysis of tumor-infiltrating lymphocytes

Fresh tumor samples obtained during surgery were immediately processed using enzymatic digestion to isolate single-cell suspensions of TILs, which were then sorted on a FACSAria II. The scRNA-seq approach utilized in this project was conducted according to the Chromium™ Single Cell 3′ v2 protocol provided by 10× Genomics. Approximately 10,000 sorted CD45+ cells were sampled for each experiment. After the initial processing, data cleaning and merging steps, 28,820 cells were analyzed. The Cell Ranger™ analysis pipelines were able to process the raw BCL dataset generated by Illumina. R and Python were the two major programming languages used in this project. Crucial open access packages, including “Seurat (R, version 3.0)” [8 (link),9 (link)], “Monocle 3 (R)” [10 (link),11 (link),12 (link)], and “CellPhoneDB (python, version 2.0)” [13 (link)], were obtained online. Briefly, dying cells, cell doublets and low-quality cells were first removed. The clean data from the three datasets were then merged and transformed. For dimensionality reduction, both the UMAP [14 (link)] and t-SNE [15 ] methods were applied. A built-in graph-based clustering analysis from “Seurat” was used to determine clusters in an unbiased manner. Clusters were annotated according to differentially expressed genes.