met-scDRS tutorial

met-scDRS is a statistical tool for finding disease associated cells and genes in single cell methylome data. This tutorial will illustrate how we used met-scDRS to find:

• Major depression disorder (MDD) associated cells in GSE215353 methylome atlas
• prioritized gene set compared to GWAS MAGMA genes
• brain regions disease association heterogeneity

Finding MDD associated cells in GSE215353 methylome atlas

data download:

Single-Cell DNA Methylation and 3D Genome Human Brain Atlas

preprocessing:

filtered for low variance genes (5th percentile or lower)

gene_variances = pd.Series(merged_adata.X.var(axis=0), index=merged_adata.var_names)
percentile_5th = gene_variances.quantile(0.05)
variance_mask = gene_variances >= percentile_5th
merged_adata = merged_adata[:, variance_mask]

as well as taking 1-raw fraction:

merged_adata.X = 1 - merged_adata.X

outputting to h5ad file:

merged_adata.write('processed-mch.h5ad')

full script documented in:

python inverse-fraction.py \
    --fraction_path ${methylation_csv} \
    --output_path ${processed_fraction}

options:

• fraction_path: path to raw fraction csv that you would like to filter for low variance genes and inverse the fraction
• output_path: path for writing the processed fraction csv

python csv-to-h5ad.py \
    --fraction_path ${processed_fraction}
    --meta_path ${meta_data_csv}
    --h5ad_path ${output_h5ad}

options:

• fraction_path: path to processed fraction output of inverse-fraction.py
• meta_data_csv: path to cells' meta data. e.g.: transcription batch, donor
• h5ad_path: output path for converted h5ad

Computation:

scdrs compute-score \
    --h5ad-file ${h5ad_path} \
    --h5ad-species ${species} \
    --gs-file ${gs_file} \
    --gs-species human \
    --out-folder ${output_dir} \
    --flag-filter-data False \
    --flag-raw-count False \
    --n-ctrl 1000 \
    --weight_opt "inv_std" \
    --flag-return-ctrl-raw-score False \
    --flag-return-ctrl-norm-score True

options:

• h5ad-file: the processed h5ad file for with inverse methylation fraction (1-raw fraction)
• h5ad-species: which species is sequenced in h5ad, supports mouse or human
• gs-file: gene set file that contain the disease putative genes more info
• gs-species: which species is putative disease genes coded for, supports mouse or human
• out-folder: folder for output, create subdirectories for each gene set enclosed in gs-file
• n-ctrl: number of control gene sets to sample
• flag-return-ctrl-raw-score: if you wish met-scDRS to return raw scores for sampled control genes
• flag-return-ctrl-norm-score: if you wish met-scDRS to return normalized score for sampled control genes

Gene set prioritization:

Compute the gene wise correlation between the processed methylation level and the met-scDRS risk score here is how we computed the prioritized genes using met-scDRS risk scores and the processed methylation level. At its core, we use the spearman correlation between risk score and methylation level to prioritized genes

License

MIT