Literature · OpenMed
OpenMed GenomicDetect
Identifies and extracts genomic and cell-line entity mentions from biomedical literature and clinical texts.
What OpenMed GenomicDetect is
OpenMed GenomicDetect (OpenMed-NER-GenomicDetect-PubMed-335M) is a token classification model developed by OpenMed for named entity recognition in biomedical text. It processes scientific publications, clinical notes, and healthcare records to identify entity mentions related to genetics, genomics, and molecular biology.
Manually curating genomic entities and genetic variants across large volumes of biomedical literature is labour-intensive. By automating mention extraction, the model supports downstream applications such as literature mining for drug discovery, construction of biomedical knowledge graphs, and genetic association studies.
How it works
OpenMed GenomicDetect is built upon the BiomedNLP-BiomedBERT-large-uncased-abstract transformer architecture. It takes tokenised English biomedical text as input and outputs BIO-tagged entity predictions. Model training employed the AdamW optimiser with learning rate scheduling, with evaluation conducted via cross-validation on a held-out test set.
The model is fine-tuned on the GELLUS corpus, a dataset specifically designed for gene recognition and genetics entity extraction in molecular biology literature. On the GELLUS benchmark, the model achieves an F1 score of 0.9963, a precision of 0.9961, a recall of 0.9965, and an accuracy of 0.9991.
Key facts
- Base architecture
- BiomedNLP-BiomedBERT-large-uncased-abstract
- Task
- Token classification (named entity recognition)
- Training dataset
- GELLUS
- Optimiser
- AdamW with learning rate scheduling
- Primary language
- English
- Recommended GPU memory
- 16 GB
What you can use it for
- Genetics literature mining
- Extracts mentions of genes, genetic variants, and hereditary conditions from molecular biology literature.
- Biomedical knowledge graph construction
- Supplies structured entity annotations from unstructured clinical and scientific texts to populate knowledge graphs.
- Drug discovery workflows
- Assists in identifying relevant biological targets, chemical entities, and adverse events from medical research documents.
- Clinical text processing
- Identifies annotated biomedical entities in patient records, healthcare notes, and clinical benchmark data.
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.
Runs on On-demand GPU
Runs on our servers with on-demand compute. A first run needs time to load the model; active capacity can be reused and scales down when idle.Run this model
OpenMed GenomicDetect
Strengths and limits
Good at
- Evaluated with an F1 score of 0.9963 on the GELLUS benchmark dataset.
- Fine-tuned from a domain-adapted biomedical language model pre-trained on PubMed abstracts.
- Integrates with standard token-classification pipelines that support multiple aggregation strategies such as simple, first, average, or max.
Watch out for
- Documented supported entity labels are restricted to B-Cell-line-name and I-Cell-line-name despite the focus on genomic text mining.
- Designed exclusively for English-language scientific and clinical text.
- Requires substantial computational resources for efficient batch processing, recommending up to 16 GB of GPU memory.
Inputs and outputs
You provide
Text
Entities are labelled token by token.
You get back
Labels with confidence scores, highest first.
Licence
Check the licence before commercial use · See model card
The See model card licence carries terms worth reading before you depend on it.
Questions
Can I use OpenMed GenomicDetect commercially?
Check the licence before commercial use. The See model card licence carries terms worth reading before you depend on it.
How do I run OpenMed GenomicDetect 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 CPU only, typically in 12s, and a run costs about 28 credits. Visitors without an account get 1,000 credits a month.
What entities does this model detect?
Its documentation lists support for B-Cell-line-name and I-Cell-line-name labels, while the GELLUS training corpus targets gene names, genetic variants, and genomics-related entities.
What underlying architecture does the model employ?
It uses the BiomedNLP-BiomedBERT-large-uncased-abstract architecture, a BERT-based transformer adapted for biomedical text.
How can entity tokens be aggregated?
Predictions can be grouped using pipeline aggregation strategies including none, simple grouping of consecutive tags, or word-level strategies such as first, average, or max.
Which training dataset was used?
The model was trained on the GELLUS corpus, which contains annotations for gene recognition and molecular biology entities.
How to cite it
Maziyar Panahi (2025). OpenMed NER: Open-Source, Domain-Adapted State-of-the-Art Transformers for Biomedical NER Across 12 Public Datasets. arXiv:2508.01630 [cs.CL].
Written from OpenMed GenomicDetect's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.