← Back to all models Foundation Model

scFoundation

Hao M, Gong J, Zeng X, Liu C, Guo Y, Cheng X, Wang T, Ma J, Zhang X, Song L · 2024-06-06 · Nature Methods

Large-scale pre-trained model for single-cell transcriptomics with 100M parameters, trained on 50M+ cells.

Overview

scFoundation is a large-scale foundation model for single-cell transcriptomics with 100 million parameters, pre-trained on over 50 million human cells. It features a novel xTrimoFormer architecture designed to handle the extreme sparsity and high dimensionality of scRNA-seq data.

Publication

Large-scale foundation model on single-cell transcriptomics

DOI: 10.1038/s41592-024-02305-7

Links

📄 Read Paper 💻 GitHub

Specifications

  • ArchitecturexTrimoFormer (asymmetric encoder-decoder)
  • Parameters100M
  • Embedding Dim512
  • Pretraining Data50M+ human cells
  • ModalityscRNA-seq

Tags

Foundation Model Gene Expression scFoundation