Files for Fidelity weighting article
These files were used to create analysis and plots associated with code and article Fidelity weighting. For code and more information go to https://github.com/sanrou/fidelityweighting. 
Files are subject specific. Files are in numpy format (.npy). Use np.load(pathToFile) to load files in Python.

File types:
forwardOperatorMEEG.npy
-	Forward solution matrix with MEG and EEG channels. Bad channels have been removed.

inverseOperatorMEEG.npy
-	Inverse solution matrix with MEG and EEG channels. Bad channels have been removed.

sourceFidelities_MEEG_X.npy
-	A vector the length of sources. Gets values between 0–1 (signed). High absolute value means the source models simulated data well.
-	A metric to estimate how well a source models simulated activity after forward and inverse modeling.
-	Source fidelities values that are signed with the potential flip (1 or -1). So a negative value indicates that if you are using flips, this source should be flipped.
-	To derive weights from source fidelities, use the formula Weight = Sign x sourceFidelity^2. Be wary of using sign, as it can be more harmful than useful.
-	Source fidelities depend on the parcellation schema. Be sure to use the same parcellation schema for source identities and source fidelities.

sourceIdentities_X.npy
-	Parcellation identities vector the length of sources (integer). Parcel identity is given as a number. So two sources that have the same number belong to the same parcel. If a source is not a part of any parcel, it has identity value -1. 
-	X is parcellation schema. Parc68 is Lausanne schema. Parc2009 is Destrieux schema. Parc2011 is Yeo schema. Parc2018 is Schaefer schema. Schaefer schema is the one used in analyses.
-	Schaefer schemas 597, 775, 942 are derived from 600, 800, and 1000 where parcels which all subjects did not have a source in are removed. *_reducedParcels.npy have parcel identities that are deleted from the original sourceIdentities file to derive the reduced parcellation schema.