When to use Batch Query
Batch Query excels at candidate prioritisation tasks: "Which of these 150 kinases have published evidence linking them to therapy resistance in lung cancer?", "Which drugs on this list have reported activity against KRAS mutant tumours?", or "Which cell types in this panel show inflammatory signatures in the context of COVID-19?". The AI searches biomedical literature for each entity independently, scores relevance against your question, and returns a ranked table you can sort, filter, and download.
Filling in the form
Navigate to formaticon.cellformatica.com/query-request or click Create Batch Query from the dashboard.
Research question Required
State your scientific question clearly. The AI uses this as the central criterion to evaluate each entity in your list. The more specific and hypothesis-driven your question, the more discriminating the results.
Good example: "Which of these genes are upregulated in KRAS-mutant colorectal cancer and correlate with poor overall survival?"
Less useful: "Tell me about these genes", which is too vague to produce a meaningful ranking.
Entity type Required
Select the category that best describes the items in your list. Options include:
- Gene: use HGNC-approved gene symbols (e.g.
TP53,EGFR) - Drug: use INN or common drug names (e.g.
imatinib,pembrolizumab) - Disease: use standard terminology (e.g.
non-small cell lung cancer) - Protein: use UniProt-style names or HGNC protein names
- Cell type: use Cell Ontology terms where possible
The entity type helps the AI interpret each list item in the correct biomedical context.
Entity list Required
Enter your entities as a comma-separated list directly in the text field, or upload a CSV or TSV file.
Manual entry
Type or paste entities separated by commas, each followed by a space: TP53, BRCA1, EGFR, MYC, PTEN.
CSV/TSV upload
Upload a file where the first column contains your entity names. A column header is optional — if present, the first row is treated as a header and skipped. Additional columns are preserved and can be merged into your results later. UTF-8 encoding is required.
Information focusing terms Optional but recommended
Focusing terms are keywords or short phrases that narrow the scope of the literature search for each entity. They help the AI retrieve the most relevant evidence rather than all published content about each gene or drug.
Example: For a question about therapy resistance in lung cancer, good focusing terms might be: drug resistance, lung cancer, NSCLC, chemotherapy, targeted therapy.
Auto-generating focusing terms
Click Proceed to make focusing terms and the AI will suggest terms based on your research question. Review the suggestions and edit them as needed before proceeding.
Entering focusing terms manually
Type terms directly into the field, separated by commas. You can combine auto-generated terms with your own additions.
Scoring question Optional
The scoring question is a yes/no question used to rank entities. For each entity, the AI answers this question based on available evidence and uses the answer to compute a relevance score. If left blank, the main research question is used for scoring.
Example scoring question: "Is there published evidence that this gene promotes resistance to EGFR inhibitors in lung cancer?"
Running the query
Once you have filled in the required fields, click Submit. A confirmation modal appears showing:
- A summary of your inputs
- The estimated credit cost
Click Confirm to start processing. The query enters a queue and begins within seconds. You can navigate away — results will be available in your Batch Query History when complete.
Viewing results
Results table
The results page shows an interactive table with one row per entity. Key columns include:
- Entity name: the gene symbol, drug name, or other identifier from your list.
- Effect score: a relevance score (typically 0–1) indicating how strongly the evidence supports a connection to your research question. Higher is more relevant.
- Novelty score: indicates how underexplored the entity is relative to your question. A high novelty score with a high effect score suggests a potentially interesting but understudied candidate.
- Precision / Recall: statistical confidence metrics for the AI's evidence retrieval.
- Answer / Evidence: a brief text summary of the evidence found for each entity.
- References: PubMed IDs (PMIDs) cited as evidence. Click any PMID to open the abstract directly on PubMed.
Click any column header to sort the table. Use the search field to filter by entity name.
Scatter plot
Below the table, an interactive scatter plot visualises the results. Use the X axis and Y axis dropdowns to select any two numeric columns from your results — for example, effect score on X and novelty score on Y to identify novel high-relevance candidates in the top-right quadrant.
Click or lasso-select points on the scatter plot to filter the table to only the selected entities. This is useful for drilling down into a subset of interest.
Merging with your own data
Upload a CSV containing additional columns (e.g. expression fold-change from your own experiment) and select the merge column. The app joins your data to the results table by entity name, letting you combine Formaticon's literature evidence with your experimental data in one view.
Downloading results
Click Download CSV to export the full results table including all columns and references. The CSV uses UTF-8 encoding and is compatible with Excel and R.
Re-running a query
Go to Batch Query History. Find your previous run and click Run again. The form is pre-filled with all your previous inputs — you can edit any field before resubmitting.
Sharing results
On the results page, click Share to generate a read-only link. Anyone with the link can view and download the table without logging in. See History, Sharing & Billing for details.