Nova Best Practices

Get the most out of Nova with these tips for effective AI-assisted scientific computing.

Be Specific

Less effective:

Analyze my data

More effective:

Analyze the RNA-seq counts in stash://myusername/rnaseq/counts.csv
Identify genes with log2 fold change > 2 and adjusted p-value < 0.05
Create a volcano plot highlighting significant genes

Provide Context

Tell Nova about your domain and goals:

I'm studying protein-ligand binding. Using the docking results
in stash://myusername/docking/, identify the top 10 compounds
by binding affinity and explain why they might be good candidates.

Iterate Progressively

Build on previous results:

# Start simple
@nova Show me the first 10 rows of my dataset

# Then refine
@nova Filter to only include samples from Group A

# Then analyze
@nova Calculate mean and standard deviation for each metric

Specify Resources

For compute-intensive tasks:

@nova Run this simulation on a MEDIUM node with GPU

Ask for Explanations

@nova Explain your methodology for this analysis

@nova Why did you choose this statistical test?

Save Important Results

@nova Save these results as a PDF report to my stash

Use Agents for Repeated Tasks

If you find yourself asking similar questions:

Create an AI Agent trained on your domain
Build a knowledge base with your protocols
Share with your team for consistent methodology

Related: Conversation Guide | AI Agents

Code Generation with Nova AI Agents