CO-SPATIAL: A Nextflow pipeline to perform spatial-ATAC-RNA seq data integration

Co-spatial is a Nextflow pipeline, inspired by nf-core practices, to generate spatial ATAC-seq and spatial RNA-seq-ready counts & peaks outputs for joint spatial ATAC-RNA sequencing downstream analysis workflows.

With Nextflow, user can simultaneously process both spatial-ATACseq and spatial-RNAseq data from the same sample, this method is made possible through (Deterministic co-Barcoding in Tissue), where single cells can now contain information from both spatial transcriptomics and spatial-ATAC-seq or spatial-CUT&Tag-seq.

Requirements

Nextflow

Docker

In order to perform downstream analysis with Signac and ArchR, users can install the docker container found under container/ and modify anything needed. However, the processes are already made convenient using a built container

This pipeline requires high computing resources to run efficiently, please modify base.config if needed.

Workflows:

1. By providing a SRA accession, the data is prefetch and fasterq-dump into FASTQ files from spatial-omics experiments.
2. The downloaded files are "branched" into either spatial-RNA or spatial-ATAC for its separate processing. 2a. For sp-RNAseq data, FASTQ files are preprocessed to extract barcode A, barcode B & UMI (read2). They are then processed using ST_pipeline, where it is aligned to STAR using a reference genome & annotation. 2b. For sp-ATACseq data, FASTQ files are preprocessed and aligned to a reference genome. The count data is then converted into 10X Genomics CellRanger-ATAC format which then can be counted using cellranger-atac
3. To perform downstream analysis, Seurat is used to connected the two modals: RNA and ATAC. Samples share the same name under samplesheet.csv file.

Inputs

In order to successfully run the pipeline, you must provide a samplesheet.csv file with the following content:

id,name,experiment
SRR22561636,ME13_50um,RNA
SRR22565186,ME13_50um,ATAC

Running the pipeline

The usual run would have the following format:

nextflow run main.nf \
    -profile docker \
    --input <path/to/samplesheet.csv> \
    --fastq_out <path/to/fastq-output-dir> \
    --spatial_barcodes <path/to/spatial_barcodes.txt> \
    --ref_map <path/to/reference-genome-dir> \
    --ref_annotation <path/to/references/genes.gtf> \
    --ref_atac_genome <path/to/references/cellranger-atac> \
    --outdir <output-dir> 

Testing the pipeline

For convenience of testing, I have created a test profile to run the pipeline on a very small set of data (10-mil reads) of 2 samples where you can download using

curl -O "https://zenodo.org/records/15241549/files/SRR22561636_subset.zip?download=1"
curl -O "https://zenodo.org/records/15241549/files/SRR22565186_subset.zip?download=1"

nextflow run main.nf \
    -profile test,docker

To test the full set of these samples, please refer to:

nextflow run main.nf \
    -profile test_full,docker

Future updates

• Integrate a CUT&Tag-RNA sequencing subworkflow
• Incorporate AWSBatch / HPC option (if I have access to one)

Citations

This pipeline is inspired by the publication: Spatial epigenome-transcriptome co-profiling of mammalian tissues

Spatial epigenome-transcriptome co-profiling of mammalian tissues.

Di Zhang, Yanxiang Deng, Petra Kukanja, Eneritz Agirre, Marek Bartosovic, Mingze Dong, Cong Ma, Sai Ma, Graham Su, Shuozhen Bao, Yang Liu, Yang Xiao, Gorazd B. Rosoklija, Andrew J. Dwork, J. John Mann, Kam W. Leong, Maura Boldrini, Liya Wang, Maximilian Haeussler, Benjamin J. Raphael, Yuval Kluger, Gonçalo Castelo-Branco & Rong Fan

Nature 2023 March 15. doi: 10.1038/s41586-023-05795-1

Source code to preprocess is modified based on original repo