Using pre-built pipeline

YASS provides with a pre-built pipeline for spike sorting, which consists of five parts: preprocess, detect, cluster, make templates and deconvolute.

Preprocess

blockdiag Butterworth fi lter Standarize Whitening
Name Description
Butterworth filter Apply filtering to the n_observations x n_channels data matrix (optional)
Standarize Standarize data matrix
Whitening Compute whitening filter

See Preprocess for details.

Detect

blockdiag Threshold dete ctor PCA Whiten scores Neural Network detector Autoencoder
Name Description
Threshold detector Detect spikes using a threshold
PCA Dimensionality reduction using PCA
Whiten scores Apply whitening to PCA scores
Neural Network detector Detect spikes using a Neural Network
Autoencoder Dimensionality reduction using an autoencoder

See Detect for details.

Cluster

See Cluster for details.

Templates

See Templates for details.

Deconvolve

blockdiag Deconvolution Merge
Name Description
Deconvolution Deconvolute unclear spikes using the templates
Merge Merge all spikes to produce the final ouput

See Deconvolute for details.