Processed data used in "Multiplexed Vulnerabilities of Striatal Neurons in Human and Rodent Models of Huntington's Disease"

Processed data used in "Multiplexed Vulnerabilities of Striatal Neurons in Human and Rodent Models of Huntington's Disease"

DEG dataset:

Cell-type-specific pseudobulk differential gene expression (DGE) analysis was performed using ACTIONet and limma for sufficiently abundant cell types using age (human), sex (human), and disease (human and mouse) phenotype as design covariates and gene-wise single-cell-level variance as weights for the linear model. Genes were considered differentially expressed if they had an FDR-corrected p-value < 0.001 and an absolute log2-fold change > 0.1 for that cell type relative to the normal control group or the reference cellular sub-type. To ensure that DGE results were reproducible and robust to differences in cell type abundance, we sampled with replacement equal numbers of mice/individuals and cells per mouse/individual for each cell type and repeated the pseudobulk analysis. Lastly, we repeated the analyses using DESeq2 as the model-fitting algorithm in lieu of limma to ensure replicability across methods. In all cases, DGE results were consistent, and we used the pseudobulk limma results for all downstream analyses.

Transcriptomic distance dataset:

We computed the average gene expression vector for each subtype and calculated the pair-wise Jensen-Shannon divergence (JSD) using the philentropy R package between all subtypes. The JSD is a measure of similarity between two distributions in the interval [0, 1] with 0 denoting two identical distributions. The Jensen-Shannon distance was defined as the square root of the JSD.

Fraction of SPN subtype dataset:

After the clustering and annotations using ACTIONet, fraction of each SPN subtypes in the entire SPN populations identified.