Replogle et al. (2022) reanalysis: perturbmatch differential expression matrices and processed objects

Differential expression matrices and processed SingleCellExperiment objects derived from reanalysis of Replogle et al. (2022) genome-wide CRISPRi Perturb-seq data in K562 cells (GEO: GSE146194) using the perturbmatch R package with limma-voom weighted GLM and perturbation-matched controls. Contents: - Log-fold-change (lfc), standard error (se), z-score (lfc/se), and cosine similarity matrices for ~9,500 perturbation targets across 4 batch splits (S1-S4) - Combined singlets+multiplets matrices (ALL condition) - Root-mean-square z-score summary across perturbations - Processed SingleCellExperiment input object (~2M cells, top 3000 HVGs, fishash 0.4.1 dual-guide adjustment) - Batch variation estimates for signal-to-noise ratio calculation Produced as part of: Yeung et al. 2026, 'Joint analysis of multiply perturbed cells improves statistical power and cost efficiency in Perturb-seq'. v2 changes (2026-07-28): Added HDF5-backed counts matrix for top 3000 HVGs (replogle_top3000_counts.h5 + replogle_top3000_se.rds). Load with HDF5Array::loadHDF5SummarizedExperiment(). Removed replogle_psce_ngenesCRISPRHVG_sce_unrealized.rds (replaced by HDF5 version).