A single-nucleus transcriptomic atlas of the adult macaque superior colliculus

The dataset captures transcriptomic profiles of isolated nuclei from the SC and does not represent a developmental time series or a spatial transcriptomics dataset. Accordingly, the deposited expression matrices contain molecular profiles at the nucleus level but do not include spatial coordinates or spatial resolution information.Superior colliculus tissue was collected from the macaque midbrain, and nuclei were isolated for snRNA-seq library preparation using the DNBelab C4 platform (BGI).

Libraries were constructed following the manufacturer's protocol for droplet-based snRNA-seq and were subsequently sequenced using the DIPSEQ T1 platform (BGI). Each library contained cDNA reads for transcript quantification and oligonucleotide reads for barcode identification.Raw sequencing reads were processed using dnbc4tools (version 2.1.3) for read processing and gene expression quantification. Gene expression matrices were generated using the Macaca fascicularis genome annotation (Ensembl release 5.0.91).

For each library, both raw and filtered gene-barcode matrices are provided in the compressed Matrix Market format. In these matrices, rows correspond to annotated genes, columns correspond to molecular barcodes representing individual nuclei, and matrix values represent UMI (unique molecular identifier) counts. Raw matrices contain all barcodes detected by the upstream processing pipeline, whereas filtered matrices contain only those barcodes retained after initial filtering.The processed snRNA-seq dataset was generated using R-based single-cell analysis workflows.

The preprocessing pipeline included ambient RNA contamination correction using SoupX, integration of multiple sequencing libraries, quality control filtering, and doublet detection. Dimensionality reduction, clustering, and cell-type annotation were performed using Seurat-based workflows, and potential technical variation associated with individual animals was mitigated using Harmony-based integration. The final cell annotations comprise seven major cell classes and 61 transcriptomic subclasses.

These major cell classes include glutamatergic neurons (GLUT), GABAergic neurons (GABA), oligodendrocytes (Oligo), oligodendrocyte precursor cells (OPC), astrocytes (Astro), microglia and border-associated macrophages (Micro_BAMs), and endothelial and vascular-associated cells (Endo_Vas).The deposited data are organized into four components: raw sequencing data (fastq), processed gene expression matrices (count_matrix), processed transcriptomic data (processed_data), and analysis scripts (code).

The fastq directory contains paired-end FASTQ files from each library, including cDNA reads, oligonucleotide reads, and sequencing quality reports. The count_matrix directory contains per-library raw and filtered sparse gene-barcode matrices together with processing reports. The processed_data directory contains the final Seurat object (SC_snRNAseq_final.rds), the nucleus-level metadata (cell_annotation_metadata.csv), and the marker-gene tables supporting cell-class and subclass annotations.

The Seurat object stores the processed expression data, metadata, dimensional reductions, clustering results, and curated cell-type annotations, and can be accessed using the Seurat package in R. The code directory contains analysis Jupyter notebooks (*.ipynb) and the R package environment specification (renv.lock) required for computational reproducibility.The nucleus-level metadata table contains one row per retained nucleus and includes nucleus identifiers, animal and library identifiers, quality-control measurements, doublet detection results, clustering information, and final cell-class and subclass annotations.

Animal-level and library-level metadata are provided separately. No expression imputation or artificial reconstruction was applied to the deposited gene expression matrices. Ambient RNA contamination correction was performed during preprocessing using SoupX, and the resulting corrected expression values were retained in the processed Seurat object.This dataset provides a molecular reference for investigating the cellular diversity, molecular organization, and evolutionary comparison of the primate superior colliculus.

The deposited resources can support diverse applications, including comparative transcriptomics, cross-species cell-type mapping, marker-gene discovery, and integration with existing human, macaque, and other mammalian brain atlases.