Connection strategies

ConnectomeMossyGlomerulus

The algorithm selects one mossy fiber within the \(x\_length \times y\_length \times z\_size\) \(\mu m\) box surrounding each glomerulus. Here z_size is the size of the partition where the glomeruli are (i.e. no limit). This selection is random and performed with a truncated exponential distribution. If no mossy_fiber could be found in the box surrounding one glomerulus, the closest mossy_fiber is chosen.

  • x_length: Length of the box along the x axis surrounding the glomerulus in which the mossy fiber can be connected.

  • y_length: Length of the box along the y axis surrounding the glomerulus in which the mossy fiber can be connected.

Note

Since the placement of mossy fibers and glomerulus is uniform within their partition, the convergence and divergence ratios correspond to the density ratios.

ConnectomeGlomerulusGolgi

The algorithm selects all glomeruli within the sphere surrounding each Golgi cell soma. For each unique Golgi cell to connect, the tip of one of its basal dendrites is selected. This selection is random and performed with a truncated exponential distribution based on the distance between the tip of each dendrite and the glomerulus to connect.

  • radius: Radius of the sphere to filter the glomeruli within it.

glomerulus_to_golgi:
  strategy: cerebellar_models.connectome.glomerulus_golgi.ConnectomeGlomerulusGolgi
  presynaptic:
    cell_types:
      - glomerulus
  postsynaptic:
    cell_types:
      - golgi_cell
    morphology_labels:
      - basal_dendrites
  radius: 50

Note

As for the glomeruli and Golgi cells, the presynaptic cell of this connection is supposed to have no morphology attached, while the postsynaptic cell must have one.

ConnectomeGlomerulusGranule

The algorithm selects n unique glomeruli within the sphere surrounding each granule cell soma, where n is equal to the convergence. Moreover, for each granule cell, each selected glomerulus should be connected to a different presynaptic cell: the presynaptic sources. The n unique presynaptic sources, their connected glomerulus, and the granule cell dendrite they will target are all randomly chosen. If not enough glomeruli could be found in the sphere surrounding the granule cell soma, the closest glomerulus connected to one of the not yet selected mossy fibers is chosen.

  • radius: Radius of the sphere to filter the presynaptic source within it.

  • convergence: Number of glomeruli per granule cell.

  • pre_glom_strats: List of ConnectionStrategy that links the presynaptic source to the glomeruli.

  • pre_cell_types: List of CellType of the presynaptic source, used as references.

glomerulus_to_granule:
  strategy: cerebellar_models.connectome.glomerulus_granule.ConnectomeGlomerulusGranule
  presynaptic:
    cell_types:
      - glomerulus
  postsynaptic:
    cell_types:
      - granule_cell
    morphology_labels:
      - dendrites
  pre_glom_strats:
      - mossy_fibers_to_glomerulus
  pre_cell_types:
      - mossy_fibers
  radius: 40
  convergence: 4

Note

As for the glomeruli and granule cells, the presynaptic cell of this connection is supposed to have no morphology attached, while the postsynaptic cell must have one.

ConnectomeGolgiGlomerulus

This connection strategy links a Golgi cell to all postsynaptic cells connected to a glomerulus: the postsynaptic targets. The connections between the glomerulus and the postsynaptic targets are configured using a list of reference strategies.

The algorithm selects here the closest glomeruli (maximum divergence) that are within a sphere surrounding each Golgi cell. For each unique glomerulus selected, the tip of an axon’s branch from the Golgi cell is randomly selected. All postsynaptic targets that are connected to the selected glomerulus with the reference strategies, are connected to the selected Golgi cell. The target points of the postsynaptic targets (i.e. dendrite selected) are copied from the reference strategies.

  • divergence: Number of glomeruli per Golgi cell.

  • radius: Radius of the sphere to filter the glomeruli within it.

  • glom_post_strats: List of ConnectionStrategy that links the glomeruli to the postsynaptic targets.

  • glom_cell_types: List of CellType of the glomeruli, used as references

golgi_to_granule:
  strategy: cerebellar_models.connectome.golgi_glomerulus.ConnectomeGolgiGlomerulus
  presynaptic:
    cell_types:
      - golgi_cell
    morphology_labels:
      - axon
  postsynaptic:
    cell_types:
      - granule_cell
  glom_post_strats:
      - glomerulus_to_granule
  glom_cell_types:
      - glomerulus
  radius: 150
  divergence: 40

Note

Note that here the Golgi cell is directly connected to the postsynaptic targets and not to the glomerulus. The latter serves only as a intermediate point to look for in the reference strategy.

Note

This postsynaptic targets are usually granule cells and unipolar brush cells.

VoxelIntersection

See bsb documentation.

ConnectomeUBCGlomerulus

The algorithm selects one UBC within the sphere surrounding each UBC glomerulus. This selection is random and performed with a truncated exponential distribution. If no UBC could be found within the sphere surrounding one UBC glomerulus, the closest UBC is chosen.

  • radius: Radius of the sphere to filter the UBC within it.

ConnectomeGlomerulusUBC

The algorithm connects a population of UBC to a list of glomeruli populations (for instance mf or ubc based glomeruli), maintaining their relative indegree ratios. Glomeruli are selected within the sphere surrounding each UBC chunk. For each UBC, the algorithm selects the closest unconnected glomerulus (if possible) to limit the number of UBCs per glomerulus.

  • radius: Radius of the sphere to filter the glomeruli within it.

  • ratios_ubc: Positive relative ratios of indegree for the different types of glomeruli. The ratios will be normalized so that their sum equals to 1.

glomerulus_to_ubc:
    strategy: cerebellar_models.connectome.glomerulus_ubc.ConnectomeGlomerulus_to_UBC
    presynaptic:
      cell_types:
        - ubc_glomerulus
        - glomerulus
    postsynaptic:
      cell_types:
        - unipolar_brush_cell
    ratios_ubc:
      ubc_glomerulus: 1. # will be interpreted as 1 / (1+2)
      glomerulus: 2.  # will be interpreted as 2 / (1+2)
    radius: 50

ConnectomeIO_MLI

The algorithm consists in a connection strategy to connect IO cells, via PC, to the molecular layer interneurons (MLI). Specifically, all the MLI connected to a PC will also receive input from the IO connected to that PC.

  • presynaptic: Define the presynaptic objects to consider. In particular, the specific cell_types can be given;

  • postsynaptic: Define the postsynaptic objects to consider. In particular, the specific cell_types can be given;

  • mli_pc_connectivity: define connectvity to be considered in the MLI subset connected to the PC;

  • io_pc_connectivity: define connectvity to be considered in the IO connected to the PC;

  • pre_cell_pc: PC population reference.

io_to_mli:
  strategy: cerebellar_models.connectome.io_molecular.ConnectomeIO_MLI
  presynaptic:
    cell_types:
      - io
  postsynaptic:
    cell_types:
      - basket_cell
      - stellate_cell
  mli_pc_connectivity:
    - basket_to_purkinje
    - stellate_to_purkinje
  io_pc_connectivity: io_to_purkinje
  pre_cell_pc: purkinje_cell

FixedIndegree

See bsb documentation.

AllToAll

See bsb documentation.

DuplicateSynapses

BSB AfterConnectivityHook to override the number of synapses per connection pair following a scipy stats function, implemented as a random distribution.