Add network interface

src/miv_simulator/config.py

@@ -246,7 +246,6 @@ class ParametricSurface(BaseModel):
 
 class CellType(BaseModel):
     template: str
-    mechanism: Optional[Dict]
     synapses: Dict[
         Literal["density"],
         Dict[
@@ -260,6 +259,7 @@ class CellType(BaseModel):
             ],
         ],
     ]
+    mechanism: Optional[Dict] = None
 
 
 CellTypes = Dict[PopulationsLiteral, CellType]

src/miv_simulator/interface/network.py

@@ -0,0 +1,104 @@
+import os
+
+from pydantic import BaseModel, Field, ConfigDict
+from machinable import Interface, get
+from miv_simulator.config import Config
+
+
+class Network(Interface):
+    class Config(BaseModel):
+        model_config = ConfigDict(extra="forbid")
+
+        config_filepath: str = Field("???")
+        mechanisms_path: str = ("./mechanisms",)
+        template_path: str = ("./templates",)
+        morphology_path: str = "./morphology"
+
+    def launch(self):
+        config = Config.from_yaml(self.config.config_filepath)
+
+        self.h5_types = get(
+            "miv_simulator.interface.h5_types",
+            [
+                {
+                    "projections": config.projections,
+                    "cell_distributions": config.cell_distributions,
+                },
+            ],
+        ).launch()
+
+        self.network = get(
+            "miv_simulator.interface.network_architecture",
+            [
+                {
+                    "filepath": self.h5_types.output_filepath,
+                    "cell_distributions": self.h5_types.config.cell_distributions,
+                    "layer_extents": config.layer_extents,
+                },
+            ],
+            uses=self.h5_types,
+        ).launch()
+
+        self.distances = self.network.measure_distances().launch()
+
+        self.synapse_forest = {
+            population: self.network.generate_synapse_forest(
+                {
+                    "population": population,
+                    "morphology": os.path.join(
+                        self.config.morphology_path, f"{population}.swc"
+                    ),
+                },
+                uses=self.distances,
+            ).launch()
+            for population in config.synapses
+        }
+
+        self.synapses = {
+            population: self.network.distribute_synapses(
+                {
+                    "forest_filepath": self.synapse_forest[
+                        population
+                    ].output_filepath,
+                    "cell_types": config.cell_types,
+                    "population": population,
+                    "distribution": "poisson",
+                    "mechanisms_path": self.config.mechanisms_path,
+                    "template_path": self.config.template_path,
+                    "io_size": 1,
+                    "write_size": 0,
+                },
+                uses=list(self.synapse_forest.values()),
+            ).launch()
+            for population in config.synapses
+        }
+
+        self.connections = {
+            population: self.network.generate_connections(
+                {
+                    "synapses": config.synapses,
+                    "forest_filepath": self.synapses[
+                        population
+                    ].output_filepath,
+                    "axon_extents": config.axon_extents,
+                    "io_size": 1,
+                    "cache_size": 20,
+                    "write_size": 100,
+                },
+                uses=list(self.synapses.values()),
+            ).launch()
+            for population in config.synapses
+        }
+
+        self.neural_h5 = get(
+            "miv_simulator.interface.neuroh5_graph",
+            uses=[
+                self.network,
+                self.distances,
+                *self.synapse_forest.values(),
+                *self.synapses.values(),
+                *self.connections.values(),
+            ],
+        ).launch()
+
+        return self