GenomicsCPU3 credits

scGPT — Single-Cell Generative Pretrained Transformer

Single-cell embeddings and marker gene identification

scGPT is a transformer pretrained on 33M cells. SciRouter exposes it for 128-dim single-cell embeddings and per-cluster marker gene identification.

$0.03
per API call
3
credits per call
/v1/singlecell/embed
API endpoint

Features

128-dim single-cell embeddings
Per-cluster marker gene identification
L2-normalized vectors ready for UMAP
Up to 100 cells per call

Quick Start

scGPT — Python Examplepython
import requests

embed = requests.post(
    "https://scirouter.ai/v1/singlecell/embed",
    json={"expression_matrix": [[...]], "gene_names": [...], "dim": 128},
    headers={"Authorization": "Bearer sk-sci-your-key"}
).json()["data"]["embeddings"]

Use Cases

1

Clustering with UMAP + Louvain

2

Cell similarity search

3

Differential expression analysis

Learn More

scGPT vs Geneformer vs scFoundation: Single-Cell Foundation Models Compared

12 min read

Related Tools

Geneformer

Zero-shot cell-type annotation from expression matrices

$0.05/call

Protein Embeddings

Generate protein language model embeddings for ML tasks

$0.02/call

Start Using scGPT

500 free credits every month. No credit card required.

Back to all tools