​​Sequencing data related to the MADM_FlpO model​

We developed a genetic mouse model called combinatorial Mosaic Analysis with Triple Markers (coMA3M) to investigate the temporal patterning of glioblastoma, a highly aggressive form of brain cancer.Sequencing data was derived from the coMA3M mouse model at various time points. Tissue preparation for SnRNA-seqAt each designated time point, 3–4 MADM<sub>hGFAP>FlpO</sub> tumor-bearing mice were euthanized by cervical dislocation: P5 (n = 3), P10 (n = 4), P20 (n = 4), P30 (n = 4), P60 (n = 3), and P90 (n = 3). Age-matched MADM<sub>hGFAP>FlpO</sub> wild-type (WT) mice were included as controls at P10 (n = 4) and P60 (n = 3). Brain tissues were rapidly dissected, and the piriform cortex—one of the primary tumor-prone regions in this model—was isolated. Approximately 40 mg of tissue per mouse was pooled and transferred into a Dounce homogenizer containing 2 mL of 1× Homogenization Buffer Unstable Solution (HBUS), composed of 242.7 mM sucrose, 0.1 mM EDTA, 0.3% NP-40, 24.3 mM KCl, 4.85 mM MgCl₂, 19.4 mM Tris-KOH (pH 7.8), 0.5 mM spermidine, 0.15 mM spermine, 1 mM DTT, and 1% protease inhibitor cocktail.Homogenization was initially performed using pestle A (10 strokes), followed by pestle B (~20 strokes) until nuclei were visibly released. The homogenate was filtered through a 70 µm cell strainer into a 2 mL Eppendorf tube and centrifuged at 350 × g for 5 minutes at 4°C. The supernatant was discarded, and the pellet was resuspended in 1× HBUS and transferred into a fresh 2 mL tube.For density gradient separation, an equal volume of 50% iodixanol solution (25 mM KCl, 5 mM MgCl₂, 20 mM Tricine-KOH, 50% iodixanol) was added to the nuclei suspension and mixed (final volume: 800 µL). A layered gradient was then prepared by sequentially adding 600 µL of 30% iodixanol (prepared by mixing 50% iodixanol with 1× HBUS at a 1.5:1 ratio) beneath the cell mixture, followed by 600 µL of 40% iodixanol (mixed at a 4:1 ratio of 50% iodixanol to 1× HBUS) at the bottom layer. The gradient was centrifuged at 3,000 × g for 20 minutes at 4°C in a horizontal swing-bucket rotor.After centrifugation, nuclei were collected and resuspended in D-PBS supplemented with 0.1 U/µL RNase inhibitor and 0.04% BSA. Single-nucleus RNA-seq libraries were prepared using the 10x Genomics Chromium Next GEM Single Cell 3′ Reagent Kit v3.1 (Dual Index), following the manufacturer’s instructions (User Manual: CG000315_ChromiumNextGEMSingleCell3’_GeneExpression_v3.1(DualIndex)_RevF).scRNA/SnRNA-seq Data Processing Raw FASTQ files were aligned to the Mus musculus reference genome (mm10) using CellRanger (v7.0.0), resulting in a cell-by-UMI count matrix. Doublets were detected and removed with DoubletFinder2 (v2.0.3). To ensure data quality, only cells meeting the following filtering criteria were retained: (i) expression of at least 300 genes, (ii) a minimum of 500 total UMI counts, and (iii) mitochondrial transcript content not exceeding 5%. Additionally, cells exhibiting over 5% expression of hemoglobin genes were classified as blood contaminants and excluded. Genes expressed in fewer than three cells were filtered out from further analysis.Subsequent data processing and analysis were conducted using the Seurat3 R package (v4.4.0). For each sample, the filtered expression matrix underwent normalization and variance stabilization via the SCTransform method with default parameters. Dimensionality reduction was performed through principal component analysis (PCA), followed by construction of a shared nearest neighbor (SNN) graph based on the top principal components. Unsupervised clustering was implemented using the Louvain algorithm, and Uniform Manifold Approximation and Projection (UMAP) was used for visualization.