Analysis of Single Cell RNASeq¶

Today it is possible to obtain genome-wide transcriptome data from single cells using high-throughput sequencing (scRNA-seq). scRNA-seq is a new technology, first publication in 2009. scRNA-seq measures the distribution of expression levels for each gene across a population of cells scRNA-seq allows to study new biological questions in which cell-specific changes in transcriptome are important, e.g. cell type identification, heterogeneity of cell responses, stochasticity of gene expression, inference of gene regulatory networks across the cells. Currently, there are several different protocols in use, e.g. SMART-seq2, CELL-seq and Drop-seq. There are also commercial platforms available, including the Fluidigm C1, Wafergen ICELL8 and the 10X Genomics Chromium.

Cell Ranger¶

The data analysis pipeline of single cell RNASeq starts with the creation of a DGE matrix, which contains gene counts in each cell, from the raw sequencing files. Cell Ranger is a set of analysis pipelines that process Chromium Single Cell 3’ RNA-seq output to align reads, generate gene-cell matrices and perform clustering and gene expression analysis.

From the Cell Ranger manual:

Cell Ranger includes four main gene expression pipelines:

cellranger mkfastq wraps Illumina's bcl2fastq to correctly demultiplex Chromium-prepared sequencing samples and to convert barcode and read data to FASTQ files.
cellranger count takes FASTQ files from cellranger mkfastq and performs alignment, filtering, and UMI counting. It uses the Chromium cellular barcodes to generate gene-barcode matrices and perform clustering and gene expression analysis. count can take input from multiple sequencing runs on the same library.
cellranger aggr aggregates outputs from multiple runs of cellranger count, normalizing those runs to the same sequencing depth and then recomputing the gene-barcode matrices and analysis on the combined data. aggr can be used to combine data from multiple samples into an experiment-wide gene-barcode matrix and analysis.
cellranger reanalyze takes gene-barcode matrices produced by cellranger count or cellranger aggr and reruns the dimensionality reduction, clustering, and gene expression algorithms using tunable parameter settings.

These pipelines combine Chromium-specific algorithms with the widely used RNA-seq aligner STAR. Output is delivered in standard BAM, MEX, CSV, HDF5 and HTML formats that are augmented with cellular information.

10x pipelines require sequencer FASTQs (with embedded barcodes) as input. The location of the 10x barcode varies depending on product and reagent version.

Single Cell Gene Expression Datasets

We download Chromium Demonstration (v3 Chemistry) > Cell Ranger 3.0.0 > 10k PBMCs from a Healthy Donor (v3 chemistry) and analyze this data using CellRanger on HTC cluster.

Description of this dataset:

Peripheral blood mononuclear cells (PBMCs) from a healthy donor (the same cells were used to generate pbmc_1k_v2, pbmc_10k_v3). PBMCs are primary cells with relatively small amounts of RNA (~1pg RNA/cell).

11,769 cells detected
Sequenced on Illumina NovaSeq with approximately 54,000 reads per cell
28bp read1 (16bp Chromium barcode and 12bp UMI), 91bp read2 (transcript), and 8bp I7 sample barcode
run with --expect-cells=10000

wget http://s3-us-west-2.amazonaws.com/10x.files/samples/cell-exp/3.0.0/pbmc_10k_v3/pbmc_10k_v3_fastqs.tar
tar xvf pbmc_10k_v3_fastqs.tar

Check the data:

zcat pbmc_10k_v3_S1_L001_I1_001.fastq.gz |head -n 20

This is 8bp I7 sample barcode.

zcat pbmc8k_S1_L007_R1_001.fastq.gz |head -n 20

This is 28bp read1 (16bp Chromium barcode and 12bp UMI).

zcat pbmc_10k_v3_S1_L001_R2_001.fastq.gz |head -n 20

This is 91bp read2 (transcript).

To analyze the dataset, submit the following job to HTC cluster.

#!/bin/bash
#
#SBATCH -t 3-00:00 # Runtime in D-HH:MM
#SBATCH --job-name=pbmc10k_cellranger
#SBATCH -c 16
#SBATCH --mem=120g

module load cellranger/3.0.2

cellranger count --id pbmc_10k_v3 \
    --fastqs pbmc_10k_v3_fastqs \
    --transcriptome=/bgfs/genomics/refs/CellRanger/refdata-cellranger-GRCh38-3.0.0 \
    --localcores=16 \
    --localmem=119 #--jobmode=slurm --maxjobs=8

--localcores=$SLURM_CPUS_PER_TASK used the allocated cores to perform computation. If your sample is large, you can use slurm to submit extra jobs --jobmode=slurm --maxjobs=8. This will submit extra 4 core jobs, and maximum number of jobs is 8.

If you get raw sequence data, an extra step mkfastq is required. You can use the following job template.

#!/bin/bash
#
#SBATCH -N 1 # Ensure that all cores are on one machine
#SBATCH -t 2-00:00 # Runtime in D-HH:MM
#SBATCH --job-name=cellrangermkfatsq

#SBATCH --cpus-per-task=4 # Request that ncpus be allocated per process.
#SBATCH --mem=60g # Memory pool for all cores (see also --mem-per-cpu)

module load cellranger/3.0.2

cellranger mkfastq --id=sampleID \
         --run=/path/to/your/sequence/data/ \
         --samplesheet=/path/to/your/sequence/data/input_samplesheet.csv \
         --localcores=$SLURM_CPUS_PER_TASK \
         --localmem=59

Under pbmc_10k_v3/outs, there is file named web_summary.html, which is a summary HTML file containing summary metrics and automated secondary analysis results. The pbmc_10k_v3/outs/raw_feature_bc_matrix contains every barcode from fixed list of known-good barcode sequences. This includes background and non-cellular barcodes. The pbmc_10k_v3/outs/filtered_feature_bc_matrix cntains only detected cellular barcodes.

Secondary Analysis in R¶

For 10x genomics single cell RNASeq data, your best option is definitely processing the raw data using the 10x Genomics free proprietary software Cell Ranger. It uses STAR to map. However, I would not trust Cell Rangers analysis beyond QC readouts, after generating the raw and filtered UMI matrices. The rest of the analysis is applied on this matrix file. QC determines which cells to exclude from downstream analysis because of various reasons like the suspicion of doublets or cellular stress. Normalization and scaling are then performed to compensate for the sparsity of data because of the low mRNA capture rate. Then, dimension reduction is done based on the most differentially expressed genes. Finally, if done correctly, visualization of the data will result in plots showing the relatedness of each cell to its neighbor in two- or three-dimensional space. You can process the unfiltered matrix with an R Bioconductor package like Seurat or Monocle 2. Both are very easy to use, were designed to be compatible with data from droplet devices like 10X, and can give you more reliable results and control over your workflow than CellRanger will.

Seurat¶

You can use RStudio server on ondemand to perform Seurat analysis. Point firefox or Chrome web browser to ondemand.htc.crc.pitt.edu. click Interactive Apps > RStudio Server, choose R version 3.6.0.

Open Rmarkdown file /bgfs/genomics/fangping/scRNASeq/pbmc10k_tutorial.Rmd, and follow the steps. You can get the following results.

Scanpy¶

Scanpy is a scalable toolkit for analyzing single-cell gene expression data. It includes preprocessing, visualization, clustering, pseudotime and trajectory inference and differential expression testing. The Python-based implementation efficiently deals with datasets of more than one million cells.

You can use jupyter notebook on ondemand to run interactive sessions. I show you how to install local python packages in virtual environment and generate ipykernel to be used on jupyter notebook on ondemand.

Logon HTC login node, load the proper python module python/bioconda-3.7-2019.03

module load python/bioconda-3.7-2019.03

conda create -n scanpy python=3.6

source activate scanpy

conda install seaborn scikit-learn statsmodels numba pytables

conda install -c conda-forge python-igraph louvain

pip install scanpy

conda install ipykernel

environment location: /ihome/fmu/fmu/.conda/envs/scanpy

python -m ipykernel install --user --name scanpy

(scanpy) [fmu@login0b ~]$ python -m ipykernel install --user --name scanpy

Installed kernelspec scanpy in /ihome/fmu/fmu/.local/share/jupyter/kernels/scanpy

Copy /bgfs/genomics/fangping/scRNASeq/pbmc10k.ipynb to your home directory.

Logon ondemand ondemand.htc.crc.pitt.edu, Interactive Apps > Jupyter Notebook

Select Python version bioconda 3.7, choose Number of hours and Number of cores, click Launch

Connect to jupyter notebook and open pbmc10k.ipynb, click Kernel > Change kernel > scanpy to activate the scanpy ipykernel.

Follow the steps. You can get the following results.