cellranger mkfastq

For details, check the tutorial from 10x Genoimcs.

On Odyssey computing cluster:

module load bcl2fastq2
cellranger mkfastq --id=test \
                   --run=/path/to/the/run/folder \
                   --csv=test.csv \
                   --jobmode=local \
                   --localmem=40 \
                   --localcores=12

test.csv is a comma seprated file with three columns:

Lane,Sample,Index
1,test_sample,SI-GA-A3

cellranger count

After cellranger mkfastq, we are ready to align the fastqs to the reference genome and count how many reads per gene per cell. These steps are wraped in cellranger count command.

cellranger count --id=sample345 \
                   --transcriptome=/opt/refdata-cellranger-GRCh38-3.0.0 \
                   --fastqs=/home/test/outs/fastq_path/HAWT7ADXX/test_sample/ \
                   --sample=mysample \
                   --expect-cells=6000

What does the output of cellranger count look like?

In the sample345 folder there is an outs folder, and you will find the files Seurat works with in the filtered_feature_bc_matrix folder. There are 3 files in the folder:

ls -sh filtered_feature_bc_matrix/
total 90M
 60K barcodes.tsv.gz  300K features.tsv.gz   90M matrix.mtx.gz
 
# The `barcodes.tsv.gz` contains the cell barcode that passed the `cellranger` filter.
zcat barcodes.tsv.gz | head -5
AAACCCAAGCGCCCAT-1
AAACCCAAGGTTCCGC-1
AAACCCACAGAGTTGG-1
AAACCCACAGGTATGG-1
AAACCCACATAGTCAC-1

# how many cells (barcodes)?
zcat barcodes.tsv.gz | wc -l
11769

# The `features.tsv.gz` contains the ENSEMBLE id and gene symbol
zcat features.tsv.gz | head -5
ENSG00000243485 MIR1302-2HG     Gene Expression
ENSG00000237613 FAM138A Gene Expression
ENSG00000186092 OR4F5   Gene Expression
ENSG00000238009 AL627309.1      Gene Expression
ENSG00000239945 AL627309.3      Gene Expression

## how many genes?
zcat features.tsv.gz | wc -l
33538


# matrix.mtx.gz is a sparse matrix which contains the non-zero counts
zcat matrix.mtx.gz | head -10
%%MatrixMarket matrix coordinate integer general
%metadata_json: {"format_version": 2, "software_version": "3.0.0"}
33538 11769 24825783
33509 1 1
33506 1 4
33504 1 2
33503 1 10
33502 1 5
33500 1 20
33499 1 9

Most of the entries in the final gene x cell count matrix are zeros. Sparse matrix efficiently save the disk space by only recording the non-zero entries.

You see the dimension of the matrix is 33538 x 11769 and the number of non-zero entries is 24825783

e.g. for the subsequent two rows in the sparse matrix:

33509 1 is the index of the row (gene) and column(cell) of that non-zero entry in the matrix, and 1 is the count number.

33506 1 is the index of the row and column of that non-zero entry in the matrix, and 4 is the count number.

Alevin

Paper : Alevin efficiently estimates accurate gene abundances from dscRNA-seq data

It supports Drop-seq and 10x-Chromium v1/2/3.

Tutorial

Kallisto/bustools

Paper: Modular and efficient pre-processing of single-cell RNA-seq

the bustoools commands we implemented are generic and will work with any BUS file, generated with data from any scRNA-seq technology. Distinct technology encodes barcode and UMI information differently, but the kallisto bus command can accept custom formatting rules. While the pre-processing steps for error correction and counting may need to be optimized for the distinguishing characteristics of different technologies, the modularity of the bustools based workflow makes such customization possible and easy.

Tutorial

Scumi

scumi is a flexible Python package to process fastq files generated from different single-cell RNA-sequencing (scRNA-seq) protocols to produce a gene-cell sparse expression matrix for downstream analyses, e.g., discovering cell types and inferring cell lineages.

It supports CEL-Seq2, 10x Chromium, Drop-seq, Seq-Well, CEL-Seq2, inDrops, and SPLiT-seq.