Clustering-for-scRNAseq

Pipelines and comparison for clustering approach in single cell RNA seq data. There are totally 6 approaches and 7 protocols for comparison on a single cell RNA sequencing benchmark dataset GSE118767:

• Mixture of H2228, H1975 and HCC827 human lung cancer cell lines: SRR6782112
• Mixture of H2228, H1975, A549, H838 and HCC827 human lung cancer cell lines: SRR8606521

Procedures

1. Upstream Analysis:

From cell ranger to count matrix

Raw SRA data -> fastq files -> count matrices

Follow the instructions from cell_ranger_pipelines
to transform raw .SRA files to count matrix produced by cell ranger pipelines.

2. Downstream Analysis:

Performing clustering analysis on count matrices:

1. Data preprocessing and Benchmarking: After getting raw count matrix, use the following files for data preprocessing and running the clustering algorithms:
   -> 3 cell lines
   -> 5 cell lines
   -> subsampling

   (Reference for individual methods could be looked up in methods/)

2. Compare them in two jupyter notebooks
   -> sc10x-3c
   -> sc10x-5c
   -> sc10x-3c-subsampling
   -> sc10x-5c-subsampling