# Compose Models Build a lab by combining models into a single runnable graph. **Time estimate**: 20 minutes ## Goal By the end of this tutorial, you'll be able to: - Understand how labs combine one or more models - Add models to a lab - Connect exposed inputs and outputs - Save and run the resulting lab ## What is Model Composition? Model composition allows you to: - **Connect models**: Link outputs of one model to inputs of another - **Multi-scale simulation**: Combine molecular, cellular, and tissue-level models - **Modular design**: Reuse validated models as building blocks - **Complex systems**: Model interactions between subsystems **Info:** Model composition is an advanced feature. We recommend completing the basic tutorials first. ## The Lab Editor Open or create a lab to work on its graph. ### Graph Canvas The main area where you arrange and connect models. - **Drag** models from the library onto the canvas - **Connect** ports by dragging from output to input - **Select** models to configure properties - **Pan** and **zoom** to navigate large compositions ### Model Picker Browse and search for models to add to your lab. ### Side Panel Review the selected model, connection, or lab settings. ## Building a Lab ### Step 1: Add Models 1. Search for your first model in the library 2. Drag it onto the canvas 3. Repeat for additional models ### Step 2: Understand Ports Each model has: - **Input ports** (left side): Accept data from other models - **Output ports** (right side): Provide data to other models Hover over a port to see its name and data type. ### Step 3: Connect Models 1. Click and drag from an output port 2. Drop onto a compatible input port 3. A connection line appears **Warning:** Only compatible ports can be connected (matching data types and dimensions). ### Step 4: Save the Lab 1. Give the lab a title 2. Add a short description 3. Save it so you can rerun or publish it later ## Example: Two Connected Models Create a simple lab with a source model feeding a downstream model: ### Add Models 1. Search for the first model you want to use 2. Add two instances to the canvas 3. Arrange them side by side ### Connect 1. Connect an output on the first model to a compatible input on the second 2. Verify the connection appears in the graph If your workflow needs feedback or nested labs, add those only after the basic forward path is working. ### Configure 1. Select each model and review its default parameters 2. Save the lab 3. Start a run to verify the graph is valid ### Run 1. Click **Start Run** 2. Set duration and output settings 3. Inspect the outputs in the results view ## Running Compositions ### Simulation Settings Labs usually expose the same run settings as any other run: | Setting | Description | |---------|-------------| | **Global Duration** | Total simulation time | ### Performance Larger labs may take longer. Tips: - Reduce output points - Start from a smaller graph first - Validate each added model before growing the lab ## Saving Labs 1. Click **Save** 2. Name your lab 3. Add description and tags 4. Share it with collaborators if your account has that capability Saved labs appear in your labs list. ## Best Practices 1. **Start simple**: Begin with two models before adding more 2. **Validate components**: Run individual models first 3. **Document connections**: Add notes explaining the biology 4. **Test incrementally**: Add one model at a time, verify each step 5. **Keep a clean lab title**: You will likely rerun it and revisit it later ## Troubleshooting ### Connection Not Allowed - Check data types match - Verify dimensions are compatible - Verify the source output and destination input are the right pair ### Simulation Fails - Check for numerical instability - Reduce time step - Verify parameter values are valid ### Unexpected Results - Verify connection directions - Run models individually to isolate issues ## Next Steps - [Biosimulant Library](/references/library) - Create compositions with Python