Input and Output (MATLAB)

Input

spikeInfo.mat

See Prepare the data for how to build it.

settings.json

See Prepare the data for how to prepare it.

See Change default settings for setting details.

Output

Output.mat

See Understand the output for details.

Motion.mat

This file contains a struct variable named Motion, which stores the estimated probe motion across sessions. The fields are listed below:

Field name	Type	Explanation
Linear	1 x n_session double	linear component of the motion
Constant	1 x n_session double	constant component of the motion. If the correction is rigid, it is the probe motion.
LinearScale	1 x 1 double	0.001 by default. It scales the Y position for numerical stability during motion estimation

The probe positions across sessions at a given Y position y can be computed as follows:

probe_positions = Motion.LinearScale*Motion.Linear*y + Motion.Constant;

See Non-rigid correction for details.

Waveforms.mat

This file contains a variable named waveforms_corrected, which is a n_unit x n_channel x n_sample x n_templates array. It stores the corrected waveform templates for all units across all channels.

resultIter.mat

This file contains a variable named resultIter, which is a 1 x n_iter struct array. n_iter is the number of accepted motion-correction iterations saved before the loop finished. If early stopping rejects a new motion estimate, that rejected iteration is not kept in resultIter. The fields are listed below:

Field name	Type	Explanation
FeatureNames	1 x n_feature cell array	features used in this iteration
Weights	1 x n_feature double	optimized weights in this iteration
IdxClusters	n_unit x 1 double	clustering result (cluster index for each unit) in this iteration
Motion	1 x 1 struct	estimated probe position (see Motion) in this iteration

SimilarityMatrix.mat

This file contains two variables named feature_names_all and similarity_matrix_all.

similarity_matrix_all is a n_unit x n_unit x n_feature 3D array containing all similarity scores for all pairs of units. feature_names_all is a 1 x n_feature array indexing the features in similarity_matrix_all. Some features are not used in later clustering steps, so their entries will be zero in similarity_matrix_all.

Intermediate files

These files include ClusterIndices.npy, DistanceMatrix.npy, LinkageMatrix.npy, and HDBSCAN_settings.json. They are used to communicate data between MATLAB and Python when running the HDBSCAN algorithm.