Because of my work, I had to learn some basic concepts and methods for single-cell research. I wrote this note to record it.

Overall Approach

In general, single-cell sequencing is a research method that considers the differences between cells by using cells as the unit for sequencing. Currently, what I see are mainly single-cell transcriptome sequencing, not genome sequencing. The experimental methods I have seen so far for implementing single-cell sequencing are:

1. Using flow cytometry to sort cells and then performing subsequent reactions in individual wells
2. The single-cell sequencing solution provided by 10X Genomics. Although it has a hint of flow-based principles, instead of putting each cell in an individual well, it delivers the cells into GEM droplets, adds barcodes within the droplets, and then sequences them on Illumina machines

From a principle standpoint, these two approaches differ in library construction and splitting. The former essentially involves selecting cells through flow cytometry and performing reactions and library construction individually in each well. Therefore, it treats each cell as an individual sample and mixes them for sequencing

On the other hand, the latter adopts a special library construction method using GEM-based experimental systems. Before constructing the library, it tags each cell’s nucleic acid with labels (16nt 10X Barcode and 10nt UMI), then mixes all the nucleic acids together to construct the library. In this way, the nucleic acids under one Illumina sample index actually come from different cells. Cell data is split based on the barcodes on the nucleic acids, and due to the presence of UMIs, they can be used for duplicate processing in subsequent steps

Technical Route

Flow Cytometry-Based Approach

• Since it essentially follows a classic method for constructing transcriptome libraries, but uses flow cytometry to reduce sample size to individual cells, the software/technological route under this approach is basically no different from that of single-sample methods. However, one may need to pay attention to issues such as duplicates and running efficiency

10X GEM-Based Approach

• Due to the use of a special library construction method, the most convenient way is to use the cellranger software suite provided by 10X Genomics. This software covers the entire analysis process from data demultiplexing to generating expression matrices and some downstream analyses (dimensionality reduction visualization, cell clustering, differential expression, marker selection), making it very convenient

Downstream Analysis Approach

Cell classification/clustering -> Differential expression -> Marker selection -> Association?