scQuint (single-cell quantification of introns)

Installation

pip install git+https://github.com/songlab-cal/scquint.git

If you need the VAE, do a full installation (with more complex dependencies):

pip install "scquint[vae] @ git+https://github.com/songlab-cal/scquint.git"

Intron quantification

We recommend starting from the intron/junction count matrix obtained by running STARsolo with options --soloFeatures Gene SJ.

How to run on Smart-seq data

How to run on 10X Chromium data (We note that only a small proportion of alternative splicing events, those close to the 3' end of the gene, can be reliably detected in 10X data.)

Starting from the splice junction output directory of STARsolo, scQuint can prepare the data with a few steps:

from scquint.data import load_adata_from_starsolo, add_gene_annotation, group_introns

adata = load_adata_from_starsolo("path/to/SJ_solo_outs")
adata = add_gene_annotation(adata, "path/to/gtf.gz")
adata = group_introns(adata, by="three_prime")
adata.write_h5ad("adata_spl.h5ad")

Precomputed AnnData objects are available at https://github.com/songlab-cal/scquint-analysis.

Differential splicing

The basic commands would be:

from scquint.differential_splicing import run_differential_splicing

diff_spl_intron_groups, diff_spl_introns = run_differential_splicing(adata, cell_idx_a, cell_idx_b)

See differential_splicing_example.ipynb for more details using Tabula Muris.

See differential_splicing_example_cortex.ipynb for more details using BICCN mouse primary motor cortex.

Citation

Benegas, G., Fischer, J., Song., Y.S. Robust and annotation-free analysis of alternative splicing across diverse cell types in mice. eLife 2022;11:e73520
DOI: 10.7554/eLife.73520