How the Interactive Runner Works with BMTK Configs¶

The Profiler.interactive_runner() method provides an interactive GUI for single-cell electrophysiological profiling using BMTK (Brain Modeling Toolkit) configuration files. This section describes how it reads and processes BMTK configs to populate the template dropdown with available cell models along with the asumptions made by the profiler interactive runner.

Initialization and Config Loading¶

When initializing the Profiler with a config parameter:

profiler = Profiler(config='simulation_config.json')

The profiler loads the BMTK simulation configuration file and extracts key paths for mechanisms and templates.

BMTK Config Structure¶

simulation_config.json¶

The main simulation config contains a "network" key pointing to the circuit configuration:

{
  "network": "circuit_config.json",
  "manifest": {
    "$BASE_DIR": "${configdir}"
  }
}

circuit_config.json¶

The circuit config defines the network structure and component directories:

{
  "manifest": {
    "$BASE_DIR": "${configdir}",
    "$COMPONENTS_DIR": "$BASE_DIR/components"
  },
  "components": {
    "mechanisms_dir": "$COMPONENTS_DIR/mechanisms/modfiles",
    "templates_dir": "$COMPONENTS_DIR/templates"
  },
  "networks": {
    "nodes": [
      {
        "node_types_file": "$NETWORK_DIR/bio_net_node_types.csv",
        "nodes_file": "$NETWORK_DIR/bio_net_nodes.h5"
      }
    ]
  }
}

Path Resolution¶

1. Mechanisms Directory: $COMPONENTS_DIR/mechanisms/modfiles

   • Contains NEURON mechanism files (.mod) compiled into shared libraries
   • Loaded via neuron.load_mechanisms(mechanisms_dir)

2. Templates Directory: $COMPONENTS_DIR/templates

   • Contains HOC files defining cell templates (e.g., templates.hoc)
   • All .hoc files in this directory are loaded into NEURON via h.load_file()

3. Node Types File: $NETWORK_DIR/bio_net_node_types.csv

   • CSV file with cell type properties
   • Must include a model_template column specifying NEURON template names

Template Discovery Process¶

1. Load HOC Templates: All .hoc files in templates_dir are loaded into NEURON, defining template classes like CP_Cell, FSI_Cell

2. Extract Template Names:

   • Reads node types from the CSV files specified in circuit_config.json
   • Looks for the model_template column in the node type data
   • Collects unique template names (e.g., CP_Cell, FSI_Cell)

3. Populate Dropdown: The interactive runner uses these template names to populate the "Template" dropdown menu

Node Types CSV Format¶

The node types CSV must be space-separated with this structure:

node_type_id model_type morphology pop_name model_template
100 biophysical PN  CP_Cell
101 biophysical FSI  FSI_Cell

• node_type_id: Unique identifier for the cell type
• model_type: "biophysical" for NEURON-based cells
• morphology: Path to morphology file (can be relative)
• pop_name: Population name (optional)
• model_template: Required - NEURON template class name defined in the HOC files

In [1]:

from bmtool.singlecell import Profiler

profiler = Profiler(config='simulation_config.json')
profiler.interactive_runner()

from bmtool.singlecell import Profiler profiler = Profiler(config='simulation_config.json') profiler.interactive_runner()

Warning: no DISPLAY environment variable.
--No graphics will be displayed.

loading /Users/gregglickert/Documents/GitHub/bmtool/docs/examples/notebooks/single_cell/components/templates/templates.hoc

VBox(children=(HBox(children=(Dropdown(description='Template:', layout=Layout(width='300px'), options=('CP_Cel…

Output(layout=Layout(border_bottom='1px solid #ccc', border_left='1px solid #ccc', border_right='1px solid #cc…