Dynamic Utilities
A module focused on functions and constructors for dynamic graph data.
Warning
Currently under development and constantly evolving. Use with caution.
CofluctTransform
Class for calculating time series coactivation. Based on Esfahlani et al, 2020.
Parameters:
Name  Type  Description  Default 

quantiles 
tuple

lower and upper bound of connection strength quantile to look at. 
(0, 1)

return_mat 
bool

Whether to return matrix (True) or vector (False). 
True

adjacency_axis 
int

position of the adjacency matrix, default being zero 
0

Example
transformer = CofluctTransform(quantiles=(0.95, 1), return_mat=True)
Cofluctuation function (cofluct)
Computes cofluctuation timeseries (per edge) for a nodes x timepoints matrix X. Based on https://www.pnas.org/content/early/2020/10/21/2005531117
Parameters:
Name  Type  Description  Default 

quantiles 
tuple

list of lowest/highest quantile of edge events to use [0, 1]: all events = pearson corr; [0, .05]: bottom 5% of events; [.95, 1]: top 5% of events 
(0, 1)

return_mat 
Whether to return a connectivity matrix (True) or dictionary (False). The dict edge contains cofluctuation timeseries (pairs_of_nodes x timepoints) and event timeseries. 
True

Returns:
Type  Description 

float

edge cofluctuation timeseries dict (pairs_of_nodes x timepoints) and event timeseries as dict 