Custom Nodes

Create your own reusable nodes with custom code, inputs, and outputs.

Creating a Custom Node

Custom nodes are defined as .md files in the nodes/ directory of your project.

Step 1: Define the Node

Create nodes/normalize_counts.md:

---
title: Normalize Counts
language: python
inputs:
  - id: input_0
    name: counts
outputs:
  - id: output_0
    name: normalized
---
 
import pandas as pd
from sklearn.preprocessing import StandardScaler
 
scaler = StandardScaler()
normalized = pd.DataFrame(
  scaler.fit_transform(counts.T).T,
  index=counts.index,
  columns=counts.columns
)
 
print(f"Normalized {normalized.shape[0]} genes")

Step 2: Use the Node

The custom node will appear in the nodebar under Custom Nodes. Drag it onto the canvas and connect it like any other node.

Input/Output Configuration

• inputs — Define named inputs that receive data from upstream nodes
• outputs — Define named outputs that pass data to downstream nodes
• Variables matching output names are automatically collected after execution

Supported Languages

Example: R Custom Node

Create nodes/run_deseq2.md:

---
title: DESeq2 Analysis
language: r
inputs:
  - id: input_0
    name: counts
  - id: input_1
    name: metadata
outputs:
  - id: output_0
    name: results
---
 
library(DESeq2)
dds <- DESeqDataSetFromMatrix(counts, metadata, design = ~condition)
dds <- DESeq(dds)
results <- results(dds)
print(paste("Found", sum(results$padj < 0.05), "significant genes"))

Sharing Custom Nodes

Custom nodes are stored as .md files, making them easy to:

• Share with collaborators
• Version control with Git
• Reuse across projects

That’s it for the tutorials! Check out the Tools and Databases references for more details on what the AI agent can do.