geraldmc/wang2023-arabidopsis-cold

Wang 2023 — Arabidopsis thaliana cold-stress expression

Arabidopsis slice of the phylotranscriptomic cold-stress dataset from Wang et al. (2023), staged for the iRI Virtual Lab 2026 R1-Q4 chain (the RNA-seq transfer/test side against the AtGenExpress microarray training set).

• Source paper: Wang, S., Zhang, Y., Ye, X. et al. A phylotranscriptomic dataset of angiosperm species under cold stress. Scientific Data 10, 399 (2023).
• Data deposit: figshare https://doi.org/10.6084/m9.figshare.22643245 (file At_read-count data.xlsx).

What this is — and an important caveat

Despite the figshare deposit being labelled "read-count data", these values are TMM-normalized StringTie expression (source sheet At.stringtie.gene.TMM.EXPR), not raw integer read counts. They are staged here as-is. Any consumer that expects raw counts (e.g. a variance-stabilizing transform / DESeq2 VST) must not be run on this matrix; put it on a log scale instead (the R1-Q4 integration uses log2(x + 1) before ComBat).

Contents

File	Shape	Notes
expression.parquet	37,351 genes × 12 samples	AGI locus IDs; TMM-normalized float. ~2% of rows are merged-locus IDs (AT...ANDAT...) that StringTie could not uniquely assign; they carry no single ATH1 probe and drop out at probe→AGI intersection.
samples.parquet	12 rows	timepoint_h ∈ {0, 2, 24, 168}, replicate ∈ {1, 2, 3}, condition (0 h → control, else cold_stress). 3 control / 9 cold.

Design

Leaf tissue (shoot-equivalent), ecotype Col-0 (inferred via Guo et al. 2023, not stated in the descriptor), four cold timepoints × three biological replicates.