Skip to main content
Getting the best results from Biomni starts with clear communication. This guide covers how to write effective prompts that help Biomni understand exactly what you need.

​
Basic Principles

​
Be Specific

The more specific your request, the better Biomni can help. 
Analyze my data

​
Provide Context

Tell Biomni about your research goals and any constraints. 
I'm studying drug resistance in cancer cells. Analyze the mutation
data in @variants.vcf to identify potentially druggable targets.
Focus on genes in the MAPK signaling pathway.

​
Use @ Mentions

Reference your files and preferred tools directly: 
Align @sample1_R1.fastq.gz and @sample1_R2.fastq.gz to the human
genome (GRCh38) using @STAR, then count reads per gene with @featureCounts

​
Examples

For examples of well-structured prompts, browse the curated Workflow Templates available in the app.

​
Tips for Better Results

​
Break Down Complex Tasks

Instead of one massive request, break it into logical steps: 
Run FastQC on @sample.fastq.gz and summarize the quality metrics

The QC looks good. Now align the reads to GRCh38 using STAR

Great, now count reads per gene and identify highly expressed genes

​
Specify Output Format

Tell Biomni exactly what you want:
  • “Generate a heatmap showing…”
  • “Create a CSV file with columns for…”
  • “Output a summary report including…”
  • “Save the results as a PDF figure”

​
Include Relevant Parameters

If you have specific requirements, include them: 
Align reads with the following parameters:
- Minimum mapping quality: 30
- Allow up to 2 mismatches
- Remove PCR duplicates
- Output sorted BAM file

​
Reference Previous Results

Build on earlier work in the conversation: 
Using the differentially expressed genes from the previous analysis,
perform GO enrichment analysis focusing on biological processes

​
Iterating on Results

Biomni learns from your feedback within a conversation:

​
Refining Results

The volcano plot looks good, but can you:
- Increase the font size for labels
- Highlight genes in the apoptosis pathway in red
- Add a title "Differential Expression: Treatment vs Control"

​
Correcting Course

Actually, I want to use edgeR instead of DESeq2 for this analysis
because my sample size is small

​
Asking for Alternatives

Can you show me a different visualization? Maybe a MA plot instead
of a volcano plot?

​
Getting Help

If Biomni doesn’t understand your request:
  1. Rephrase your prompt with more specific details
  2. Break down complex requests into smaller steps
  3. Provide examples of what you’re looking for
  4. Ask Biomni to explain what information it needs
I'm not sure how to phrase this - I have RNA-seq data from 3 timepoints
and want to find genes that change over time. What information do you
need from me to set up this analysis?