OmicVerse: An Agent-Enabled Unified Framework for Bulk, Single-Cell, and Spatial Transcriptomics Data Analysis

Recent advances in bulk, single-cell, and spatial transcriptomics have transformed biological discovery, yet fragmented tools hinder cross-modal integration. OmicVerse is a unified Python framework linking bulk RNA-seq, single-cell RNA-seq, and spatial transcriptomics through a standardized API and data model. It supports preprocessing, gene selection, cell type annotation, trajectory inference, batch correction, spatial mapping, and bulk-to-single-cell deconvolution with GPU–CPU acceleration. A novel Agent-based architecture enables adaptive analysis: dynamically benchmarking algorithms, optimizing pipelines, and integrating foundation models (e.g., scGPT, Geneformer, CellPLM) for embeddings, annotation, and conversational workflows. OmicVerse further offers publication-ready visualization, MOFA+ multi-omics integration, and pathway enrichment. This protocol guides installation, configuration, and execution from raw data to integrative, AI-augmented analyses, establishing a scalable standard for reproducible, cross-modal transcriptomics.