trWav2fet_()

Extract clustering features from given spike waveforms (tnWav1_spk) according to the specified feature (P.vcFet). Default feature is P.vcFet=gpca with P.nPcPerChan=1 (recommend for high-density probes such as Neuropixels probes). gpca refers to global PCA which uses the common basis across all channels. Multiple features that are supported by JRCLUST are described here.

Syntax

• [mrFet1, mrFet2, mrFet3, trWav2_spk] = trWav2fet_(tnWav1_spk, P)
• [mrFet1, mrFet2, mrFet3, trWav2_spk] = trWav2fet_(tnWav1_spk, P, nSites_spk, viSite2_spk)

Input

• tnWav1_spk: Filtered spike waveforms (nSamples_spk x nSites_spk x nSpikes: int16).
• P: Parameters struct (P)
• nSites_spk: Number of sites to extract spikes (optional, default: [])
• viSite2_spk: List of sites to use to compute local CAR (optional, default: [])

Output

• mrFet1: First feature (nSites_fet x nSpikes: single)
• mrFet2: Second feature (nSites_fet x nSpikes: single)
• mrFet3: Third feature (nSites_fet x nSpikes: single)
• trWav2_spk: Spike waveforms after subtracting the local CAR (LCAR)

Operations

• spkwav_car_(): Subtract local CAR from the spike waveforms. Local reference is selected by averaging the outer half of the sites. Total number of sites per spiking event is nSites_spk, which includes neighboring sites within P.maxDist_site_spk_um (default:75) from the peak site. Total number of sites used for feature extraction is nSites_fet, which includes neighboring sites within P.maxDist_site_um (default:50).
• Feature extraction according to the specified option (P.vcFet)
  • If P.fInterp_fet==1 and P.vcFet=={'pca', 'gpca', 'fpca'}, pca projection is performed after interpolating the waveform by 2x.

See also

wav2spk_(), detect_()