Genomics & Single-Cell · Theodoris et al., Gladstone Institutes
Geneformer
A transformer model for gene network inference and target prioritisation from single-cell transcriptomes.
What Geneformer is
Geneformer is a genomics transformer developed by Theodoris et al. at the Gladstone Institutes. It addresses gene network inference and disease target prioritisation using single-cell transcriptomic profiles.
The model allows researchers to examine gene network contexts and determine candidate genes relevant to disease phenotypes computationally, reducing the immediate need to execute physical laboratory experiments.
How it works
Geneformer is a transformer pretrained on 30 million transcriptomes. It represents expression profiles using rank-value encoding to learn gene network context.
To evaluate gene function, the model performs in silico deletion analysis on user-supplied expression matrices, calculating phenotypic shift magnitudes to rank candidate genes.
Key facts
- Version
- 2.0
- Architecture
- Transformer
- Training data
- 30 million transcriptomes
- Modality
- Single-cell transcriptomics
- GPU memory
- 40 GB
What you can use it for
- In silico gene deletion
- Simulate the deletion of specific genes in transcriptomic data to evaluate their impact on disease phenotypes computationally.
- Target prioritisation
- Prioritise candidate genes with associated shift magnitudes to guide downstream biological investigation.
- Gene network inference
- Infer gene network dynamics and functional relationships directly from single-cell expression matrices.
Run it here
No installation and no GPU of your own: fill in the inputs and it runs on our servers. Results stay in your browser unless you sign in.
Not runnable yet
This model isn't available to run right now.Run this model
Geneformer
Strengths and limits
Good at
- Pretrained on a large collection of 30 million transcriptomes.
- Identifies disease-relevant genes via computational in silico deletion without requiring physical assays.
- Employs rank-value encoding to model gene network context.
Watch out for
- Requires 40 GB of GPU memory to operate.
- Accepts expression matrices specifically formatted as h5ad files.
Inputs and outputs
You provide
Expression matrix (h5ad URL)
Genes to delete in silicooptional
You get back
Prioritised genes with shift magnitude.
Licence
Commercial use permitted · Apache-2.0
The Apache-2.0 licence allows use in commercial research and products.
Questions
Can I use Geneformer commercially?
Commercial use permitted. The Apache-2.0 licence allows use in commercial research and products.
How do I run Geneformer without installing anything?
Use the form under “Run it here” on this page: fill in the inputs and press Run model. It runs on our servers on 2× A100, typically in 7.0min, and a run costs about 2,520 credits. Visitors without an account get 1,000 credits a month.
What inputs are required to run Geneformer?
The model requires an expression matrix URL in h5ad format and the list of genes targeted for in silico deletion.
What does the model output?
It produces a ranking table containing prioritised genes alongside their calculated shift magnitudes.
What is in silico deletion analysis?
It is a computational method that removes genes in simulation to evaluate which genes contribute to a disease phenotype without laboratory testing.
How much GPU memory is needed?
Geneformer requires 40 GB of GPU memory.
Written from Geneformer's own documentation. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.