Skip to main content
The @ mention system connects your requests to specific resources. This page covers advanced features and troubleshooting beyond the basics covered in the Quickstart.

​
Mention Categories

When you type @, the dropdown organizes resources into tabs:
TabContainsExamples
AllEverything availableAll resources and files
FilesYour uploaded files and results@sample.fastq, @results.csv
DatabasesReference databases@UniProt, @NCBI, @PDB
ToolsAnalysis software@STAR, @GATK, @AlphaFold
PackagesR/Python libraries@DESeq2, @Seurat, @pandas

​
Search and Filtering

​
Fuzzy Matching

The mention search uses fuzzy matching—you don’t need exact names:
  • @deseq matches DESeq2
  • @uniprot matches UniProt
  • @sample matches all files containing “sample”

​
Keyboard Navigation

KeyAction
↑ ↓Move selection
Tab or EnterConfirm selection
EscClose without selecting
TypeFilter results

​
How Mentions Affect Execution

When you mention a resource, Biomni:
  1. Loads context: Documentation, parameters, and constraints for that resource
  2. Validates compatibility: Checks if the resource works with your data type
  3. Sets defaults: Applies recommended parameters unless you specify otherwise
  4. Handles dependencies: Automatically includes required supporting tools

​
Example: Tool Context

Align my reads using @STAR
Biomni knows:
  • STAR requires paired-end or single-end FASTQ files
  • Default parameters for RNA-seq alignment
  • Output will be BAM format
  • Splice junctions will be detected

​
Example: Database Context

Look up TP53 in @UniProt
Biomni knows:
  • Query the human TP53 protein entry
  • Return sequence, function, and annotations
  • Include known variants and diseases

​
Multiple Mentions

Chain resources in a single request: 
Align @sample_R1.fastq.gz and @sample_R2.fastq.gz using @STAR,
quantify with @featureCounts, then run differential expression
with @DESeq2 comparing treatment vs control
Biomni determines the execution order and data flow between tools.

​
File Mentions

​
Uploaded Files

Reference files you’ve uploaded: 
Analyze @experiment_data.csv

​
Result Files

Reference output from previous analyses: 
Using the @aligned_reads.bam from earlier, call variants with @GATK

​
File Patterns

Mention multiple related files: 
Process all the FASTQ files: @sample1.fastq.gz, @sample2.fastq.gz, @sample3.fastq.gz

​
Mention Scope

Files available for mention depend on your current context:
ContextAvailable Files
Quick TaskPersonal uploads only
Project TaskProject files + personal uploads
Within conversationAll above + results from current session

​
Troubleshooting

  • Check spelling (fuzzy search helps but has limits)
  • Try the category tabs to narrow down
  • Verify the resource is available in your plan
  • Refresh the page if recently uploaded
  • Ensure upload completed (check progress indicator)
  • Check if file is in current project scope
  • For results, ensure the generating task completed
  • Must select from dropdown, not just type the name
  • Ensure there’s a space before @
  • Selected mentions appear as highlighted chips
  • Multiple resources may have similar names
  • Use category tabs to find the correct one
  • Check the description shown in the dropdown

​
When Not to Use Mentions

Mentions are optional. Biomni can often infer resources from context: 
Run differential expression analysis on my RNA-seq data
Biomni will select appropriate tools (likely DESeq2 or edgeR) based on your data. Use explicit mentions when you:
  • Want a specific tool over alternatives
  • Need to reference a specific file among many
  • Want to ensure exact parameters are used