Refining the JSON Model Representation Format

[r-makarov]

Help us design better JSON format for neuronal models! ✨

We're developing a comprehensive JSON format for representing detailed biophysical neuronal models, and we need your expertise to make it as useful and intuitive as possible.

Our format is inspired by the SONATA JSON specification but extends it to handle more complex ion channel distributions found in detailed biophysical models.

The JSON representation captures three key aspects:

1. Domain-to-mechanism mapping: Which ion channels are inserted in each morphological domain.

    "domains": {
        "apic": [
            "CaDyn",
            "CaHVA",
            "CaLVA",
            "KCa",
            "Ka",
            "Km",
            "Kv",
            "Leak",
            "Na"
        ],
        ...

2. Segment groups which are further used to distribute various parameters

"groups": [
        {
            "name": "all",
            "domains": [
                "apic",
                "axon",
                "dend",
                "soma"
            ]
        },
        {
            "name": "apical",
            "domains": [
                "apic"
            ]
        },
        {
            "name": "dendritic_thin",
            "domains": [
                "dend",
                "apic"
            ],
            "select_by": "section_diam",
            "max_value": 0.8
        },
        {
            "name": "proximal_apical",
            "domains": [
                "apic"
            ],
            "select_by": "distance",
            "max_value": 260
        },
      ...
    ],

3. Parameter distributions: How parameters are distributed across groups using various functions

"params": {
        "cm": {
            "all": {
                "function": "constant",
                "parameters": {
                    "value": 2
                }
            },
            "somatic": {
                "function": "constant",
                "parameters": {
                    "value": 1
                }
            }
        },
       ...
        "gbar_CaHVA": {
            "all": {
                "function": "constant",
                "parameters": {
                    "value": 0.0
                }
            },
            "somatic": {
                "function": "constant",
                "parameters": {
                    "value": 5e-06
                }
            },
            "basal": {
                "function": "linear",
                "parameters": {
                    "slope": 1e-08,
                    "intercept": 5e-06
                }
            },
            "proximal_apical": {
                "function": "sinusoidal",
                "parameters": {
                    "amplitude": 4.923e-06,
                    "frequency": 0.008758,
                    "phase": 0.8656
                }
            },
            "distal_apical": {
                "function": "constant",
                "parameters": {
                    "value": 2e-06
                }
            }
        },
        "eca": {
            "all": {
                "function": "constant",
                "parameters": {
                    "value": 140
                }
            }
        },
       ...

You can learn more about domains, segment groups and parameter distributions in our tutorail.

Also take a look at these complete model representations:

• Park et al. 2019
• Poirazi et al. 2003

If you think this model representation format could be better structured, documented, etc. please leave your feedback here!