Skip to content

Latest commit

 

History

History
61 lines (51 loc) · 4.23 KB

pgmspecs.md

File metadata and controls

61 lines (51 loc) · 4.23 KB

Neo4j Property Graph Model

Property Graph Model

Notes:

  • All nodes have a time stamp (property name: timeStamp) indicating last update in the format YYYY-MM-DDTHH:MM:SS.
  • Segment nodes with greater than a specified number of synaptic densities are also labeled Neuron.
  • ConnectsTo weights are equal to the number of SynapsesTo relationships for a particular connection. Assuming that the ratio of pre to post for each synaptic connection can be 1:1 or 1:many (but never many:1) this is equal to number of postsynaptic densities for a connection.
  • Each neuron/segment has one SynapseSet node, which Contains all the Synapse nodes for that body. Each ConnectsTo relationship has a corresponding ConnectionSet node, which Contains all the Synapse nodes involved in that particular connection. These can be accessed via From and To relationships with the presynaptic and postsynaptic neuron, respectively
  • A neuron/segment is considered to be in an ROI if it has >= 1 presynaptic or postsynaptic density in that ROI.

Features specific to the hemibrain dataset:

  • All Segment nodes with >=2 t-bars, >=10 psds, a name, a status, or a soma are labeled Neuron.
  • The Meta node has properties preHPThreshold and postHPThreshold, which indicate the threshold at which the synapse classifier acheives high precision (rather than high recall) for presynaptic and postsynaptic densities, respectively.
  • ConnectionSet nodes have an roiInfo property that provides pre and post counts for that connection as well as preHP and postHP counts calculated using the above high-precision thresholds.
  • ConnectsTo relationships have a weightHP property that indicates the high-precision weight (number of high-precision postsynaptic densities) for the connection.

:Neuron/:Segment properties

  • pre: number of presynaptic densities
  • post: number of postsynaptic densities
  • size: size of body in voxels
  • name: name of neuron
  • type: type of neuron
  • bodyId: int64 identifier (unique per dataset)
  • status: status of neuron
  • somaLocation: 3D Cartesian location
  • somaRadius: radius of soma
  • roiInfo: string containing json map in format {"roiA":{"pre":1,"post":2},...}
  • <roi>: boolean indicating that body is located in a particular roi (if present, always true)
  • (Neurons only) clusterName: name generated using input and output ROIs for a neuron. If a neuron has greater than 10% of either its presynaptic or postsynaptic densities in an ROI, that ROI is added to the clusterName. ROIs are listed in lexicographic order within the name, and are in the format <input roi 1>.<input roi 2>...<input roi n>-<output roi 1>.<output roi 2>...<output roi n>. For example, roiA.roiB-roiB.roiC.roiD. If there is no input and/or output ROI in which a neuron has greater than 10% of synapses, this part of the name will be "none" (e.g. none-roiA for no input ROI). This property is used to quickly find neurons with similar connectivity patterns.

:Synapse properties

  • type: type of synapse
  • confidence: confidence
  • location: 3D Cartesian location (unique per dataset)
  • <roi>: boolean indicating that synapse is located in a particular roi (if present, always true)

:SkelNode properties

  • location: 3D Cartesian location
  • radius: radius
  • rowNumber: row number within swc file
  • type: type of node indicated in swc file

:ConnectsTo properties

  • weight: number of postsynaptic densities per connection
  • weightHP: number of high-precision postsynaptic densities per connection

:ConnectionSet properties

  • roiInfo: string containing json map in format {"roiA":{"pre":1,"post":2,"preHP":1,"postHP":1},...} where preHP and postHP indicate counts of high-precision synaptic densities

:Meta

  • lastDatabaseEdit: date and time of last database edit (YYYY-MM-DDTHH:MM:SS).
  • dataset: string indicating dataset name
  • totalPreCount: number of presynaptic densities in dataset
  • totalPostCount: number of postsynaptic densities in dataset
  • roiInfo: string containing json map in format {"roiA":{"pre":1,"post":2},...}
  • preHPThreshold: float (high-precision pre threshold)
  • postHPThreshold: float (high-precision post threshold)

:DataModel

  • dataModelVersion: property graph model version number for database