Scientific data analysis without the black box.
AI agent that runs human-verified data analysis recipes—and adds new steps as you ask follow-up questions. Executed locally on your machine, optionally with your own models or API keys.
Supports macOS 12+ and Windows 10/11. All releases
Human-maintained recipes, driven by conversation
Watch how a natural-language request activates a deterministic R & Python recipe, and extends dynamically when you ask a follow-up question.
can you run differential expression on counts.csv with metadata.tsv?
There's a library recipe that matches this task: “Bulk RNA-Seq Differential Expression”. Click Use Recipe in the steps above, say “run it”, or ask what it does.
run it
I ran the Bulk RNA-Seq Differential Expression recipe on counts.csv (18,420 genes across 24 samples: 12 Treated vs. 12 Control).
Key QC & DE results:
- Genes retained after low-count filtering: 14,890
- Significant DEGs (FDR < 0.05): 412 (248 up, 164 down)
- Top hits: CXCL10, ISG15, IFIT1, MX1 (interferon signaling)
check PubMed and drug target patents for these top 4 interferon genes
Appended 03_lit_patent_agent to the recipe graph and re-collated the report. Found 19 matching clinical/preclinical PubMed studies and 6 active patent filings targeting CXCL10, ISG15, IFIT1, and MX1 pathways.
1. Data Input
counts.csv + metadata.tsv · Raw Count Matrix · 18,420 × 24
2. 01_qc_norm.py
Python · Low-count filter & TMM/VST normalization
3. 02_deseq2.R
R · Wald test, apeglm shrinkage & volcano plot
4. 03_lit_patent_agent
AI Agent + Tools · PubMed, Patent, Dataset & Web Search
5. Collated Output
volcano_plot.svg, deg_table.csv, rna_seq_report.pdf