Combined topological and spatial constraints are required to capture the structure of neural connectomes

Repository for the manuscript "Combined topological and spatial constraints are required to capture the structure of neural connectomes"

The folders titled *_data (where * is fly, mouse, and human) contain the processed individual neuron information and connectome/contactome structure.

Below is a brief summary of the columns of the .csv files contained in each of these folders.

Summary of processed data

Neuron information

*_basic_neuron_info.csv

id : ids of the neurons in the publicly released data sets 

x_cm (y_cm, z_cm) : x (y, z) position of the center of mesh, nm

*_extended_neuron_info.csv

id : ids of the neurons in the publicly released data sets 

x_soma (y_soma, z_soma) : x (y, z) position of the soma, nm 

x_cm (y_cm, z_cm) : x (y, z) position of the center of mesh, nm 

degree : undirected degree of each neuron 

presynaptic_degree : pre-synaptic degree (# of post-synaptic neighbors) of each neuron 

postsynaptic_degree : post-synaptic degree (# of post-synaptic neighbors) of each neuron 

weighted_presynaptic_degree : # of relevant (established with other neurons we consider) pre-synapses of each neuron 

weighted_postsynaptic_degree : # of relevant (established with other neurons we consider) post-synapses of each neuron 

pca_1_x (pca_1_y, pca_1_z) : dominant principle component x (y, z) direction based on the mesh vertices 

evr_1 : explained variance ratio of the dominant principle component direction 

linear_span_pca_1 : linear span of the neuron along the dominant principle component direction 

n_mesh_vertices : number of mesh vertices

Note that we use the index of the row corresponding to each neuron (e.g., integers from 0 to 15731 for the human data) as connectome and contactome node labels in the .csv files below.

Connectome

*_directed_connectome.csv 

Here, each row corresponds to a pair of neurons (i,j), s.t. an edge from i to j exists 

i : pre-synaptic neuron index 

j : post-synaptic neuron index 

weight : number of synapses from neuron i to neuron j

*_weighted_connectome.csv

Here, each row corresponds to a pair of neurons **(i,j)**, i<j, s.t. an edge between i and j exists 

i : neuron index 

j : neuron index 

reciprocated : indicates whether the directed edge between i and j is reciprocated (exists in both directions)

*_connectome_edge_weights.csv

weight : number of synapses between two neurons 

count : number of pairs of neurons with an edge of a given weight

Contactome

*_connectome_edge_weights.csv

Here, each row corresponds to a pair of neurons in physical contact 

i : neuron index 

j : neuron index 

Summary of models

We include the edge probabilities obtained for each model discussed in the manuscript (ER, model_c, model_d, model_d_c, model_k, model_k_c, model k_L) as .npy files in models/*.zip, where * is fly, mouse, and human.  

As an example of getting edge probabilities in Python: 

import numpy as np

from scipy.spatial.distance import squareform

p = squareform(np.load('models/mouse/p_model_k.npy'))

There, p[i,j] corresponds to the probability of forming an edge between neurons i and j.

Example code for models 

We include the example code to obtain edge probabilities for models k/k+c—model_k.py—and k+L—model_k_L.py. Note that the attached code can be used to impose any hard edge constraint.