Single cell RNA Sequencing Analysis

Used for filtering, normalization, scaling, integration (optionally), and clustering of single or aggregated single-cell RNA-Seq datasets

The main functional blocks of sc-rna-analyze-wf.cwl workflow are shown below. For a detailed workflow structure refer to CWL Viewer.

---

First, make sure you have cwltool, Docker, git and wget tools installed, then proceed to the following steps.

❗ With the minimum required Docker configuration (4 CPU and 20GB of RAM) the approximate running time is up to 1 h.

1. Create a temporary folder and clone the current repository.

   mkdir sc_rna
   cd sc_rna
   git clone https://github.com/Barski-lab/sc-seq-analysis.git

2. Create a folder for input data. Download required input files from the Figshare either using a web browser or commands below.

   mkdir inputs
   cd inputs
   wget -O filtered_feature_bc_matrix.tar.gz https://figshare.com/ndownloader/files/34819513
   wget -O aggregation.csv https://figshare.com/ndownloader/files/34819516
   wget -O condition.csv https://figshare.com/ndownloader/files/34819519
   wget -O mouse_cell_cycle_genes.csv https://figshare.com/ndownloader/files/34822054

3. Copy the job definition file into the inputs folder.

   cp ../sc-seq-analysis/jobs/sc-rna-analyze-wf.yaml .

4. Create a folder for workflow outputs and execute sc-rna-analyze-wf.cwl workflow with sc-rna-analyze-wf.yaml job definition file.

   cd ..
   mkdir outputs
   cd outputs
   cwltool ../sc-seq-analysis/workflows/sc-rna-analyze-wf.cwl ../inputs/sc-rna-analyze-wf.yaml

Expected outputs (some of the plots and files are omitted)

Note, as the Docker image is being frequently updated, your outputs may not exactly correspond to the plots below. In order to reproduce exactly the same outputs, switch to 4819746 commit.

Clustering results can be also evaluated interactively in UCSC Cell Browser using RangeHTTPServer or any other simple HTTP server.

cd html_data
python3 -m RangeHTTPServer   # open http://localhost:8000/

Example of UCSC Cell Browser window.

Step 1. QC metrics and the results of low-quality cells removal.

Before low-quality cells removal	After low-quality cells removal

Step 2. Dimensionality reduction and evaluating confounding sources of variation.

Step 3. Cluster analysis and gene markers identification.

Example of the table with identified gene markers (top 10 rows)

resolution	cluster	feature	p_val	avg_log2FC	pct.1	pct.2	p_val_adj
0.5	0	Dpt	0	3.06828808	0.941	0.176	0
0.5	0	Col3a1	0	2.57320598	0.998	0.814	0
0.5	0	Lum	0	2.48541392	0.951	0.318	0
0.5	0	C4b	0	2.44687844	0.962	0.444	0
0.5	0	Fbn1	0	2.42990848	0.978	0.524	0
0.5	0	Clec3b	0	2.4144368	0.767	0.134	0
0.5	0	Col14a1	0	2.39398771	0.897	0.167	0
0.5	0	Sfrp1	0	2.37871487	0.852	0.304	0
0.5	0	Gsn	0	2.37083093	0.946	0.602	0