BioBERT quickstart

Access the BioBERT tool

BioBERT no longer has a form in .bloom*'s Tools panel, and chat cannot call it (see the note above). The steps below describe the BioBERT workflow so you can read existing BioBERT results. BioBERT provides named entity recognition, relation extraction, and question answering for biomedical text.

Choose your task

Named Entity Recognition (NER):

• Identify genes, diseases, drugs, and other entities
• Get entity spans and classifications
• Batch process multiple documents

Relation Extraction (RE):

• Extract relationships between entities
• Drug-disease, protein-protein, gene-disease links
• Evidence sentences for each relationship

Question Answering (QA):

• Ask questions about biomedical text
• Get extracted answers with evidence
• Support for complex biomedical queries

Enter your text

Direct text input:

Aspirin inhibits cyclooxygenase (COX) enzymes, reducing
prostaglandin synthesis. This mechanism provides anti-inflammatory
and analgesic effects. Common side effects include gastrointestinal
bleeding and increased bleeding time.

File upload:

• Plain text files (.txt)
• PDF documents (text extraction)
• CSV with text column

Configure settings

Entity types (NER):

• Genes/Proteins
• Diseases
• Chemicals/Drugs
• Species
• Mutations

Relationship types (RE):

• Drug-Disease
• Drug-Target
• Gene-Disease
• Drug-Drug Interaction

View results

NER output example:

[Aspirin]DRUG inhibits [cyclooxygenase]PROTEIN ([COX]PROTEIN)
enzymes, reducing [prostaglandin]CHEMICAL synthesis.
Side effects include [gastrointestinal bleeding]DISEASE.

RE output example:

Relation: DRUG-TARGET
Entity 1: Aspirin (Drug)
Entity 2: cyclooxygenase (Protein)
Relation type: inhibits
Confidence: 0.94

Export and integrate

Export formats:

• Annotated text (standoff format)
• CSV with entities and relationships
• JSON for programmatic access

Integration:

• Feed entities to knowledge graphs
• Use relationships for target discovery
• Support literature review workflows