Linking Microbial Genetic Signals and Single-Cell Transcriptomics to Unravel Host-microbe Interactions across Human Tissues

Overview:

We introduce scBPS to incorporate microbial GWAS summary data with human single-cell transcriptomic data for the discovery of critical cellular contexts that are highly associated with gut microbial taxon.

How to run

snakemake -s scBPS.smk \
    --cores N \
    --config zscore_file=<path_to_zscore_file> \
             scfile=<path_to_scfile> \
             outdir=<output_directory> \
             anno=<path_to_annotation_file> > log.txt 2>&1

Example

snakemake -s scBPS.smk \
    --cores 4 \
    --config zscore_file=test/test_gene_zscore.txt  \
             scfile=test/test_adata.h5ad \
             outdir=output \
             anno=test/cell.annotation.txt > log.txt 2>&1

Parameters

zscore_file: This file contains two columns and is used for storing GWAS-derived top genes and their corresponding gene weights for score calculation. Each row corresponds to a specific trait and its associated gene set. The two columns are:

• The first column TRAIT Represents the specific trait or phenotype from the GWAS analysis.
• The second column GENESET contains the associated set of genes for each trait, listed in a comma-separated format.

scfile: This file contains single-cell RNA sequencing data in .h5ad format (AnnData). It stores gene expression information across individual cells (required, size-factor-normalized and log-transformed), along with any associated annotations such as cell type and metadata.

anno: This file contains two columns and provides cell annotations for single-cell RNA sequencing data. The columns are as follows:

• The first column cell_id contains the unique cell identifier (ID) for each cell.
• The second column cell_annotation contains cell annotation information. Each unique group in this column will be treated as a distinct cell population, which will be used to calculate the BPS_AUC score and p-value. Cells that do not belong to any of the relevant or interested cell population groups should be labeled as “other” in this column.

outdir: This directory specifies the output location where all results will be saved.

--cores: The number of CPU cores to be used.

Outputs

norm_score.tsv: A tab-separated text file containing cell-wise scores for each cell across all traits.

BPS_AUC.txt: This file contains the calculated BPS_AUC scores for all pairs of cell populations and traits. The cell populations are provided by the anno parameter (specifically, the cell_annotation column), and the traits are taken from the input zscore_file. Each row corresponds to a cell population, and each column represents a specific trait.

pvalue_AUC.txt: This file contains the p-values that represent the statistical significance of the BPS_AUC scores for each cell population (as defined in the cell_annotation column from the anno parameter) across all traits from the input zscore_file. Each row corresponds to a cell population, and each column represents a specific trait.

Installation and Requirements

Before running the workflow, you need to install the required software and dependencies:

• Python 3.7+
• Snakemake
• R 4.2
• Required Python packages: scanpy, scdrs, pandas, numpy, argparse
• Install scBPS rpackage: devtools::install_github("jjlea/scBPS", subdir = "rpackage")
• Required other R packages: DelayedArray, data.table, reshape2, DelayedMatrixStats

Citation:

"Li, J., Ma, Y., Cao, Y. et al. Integrating microbial GWAS and single-cell transcriptomics reveals associations between host cell populations and the gut microbiome. Nat Microbiol 10, 1210–1226 (2025). https://doi.org/10.1038/s41564-025-01978-w"