Roberto

I have 2 NPX1 per recording. Each NPX will have 4 different xcoords. There aren´t noisy channels really. If I create the sort_analyzer from only one sorting (not the aggregated...

Ok channel selected from here are messed up. I will try to understand the precise reason later or tmrrw. https://github.com/SpikeInterface/spikeinterface/blob/d5f1481ffce3a29aabedc95920af5d091b1e1720/src/spikeinterface/core/sparsity.py#L299-L316

Somehow some channels that belong to group1 have x coord corresponding to group0. I guess that´s actually a Neo import bug.

Moved to NeuralEnsemble/python-neo#1548

I am reopening this because I'm not sure any more it is a neo or probeinterface problem test rec file here: https://www.dropbox.com/s/bsjd1ipx430dr00/20240124_171519.rec?dl=0 In this case the xcoord of group 0...

The problem is solved if I set `raw_rec.set_property("contact_vector", None)`. Doing so I directly access locations here: https://github.com/SpikeInterface/spikeinterface/blob/358e0d31df092ed4516f51f2277b2467234d6b6f/src/spikeinterface/core/baserecordingsnippets.py#L347-L367 By solved I mean that plot_unit_waveforms shows sensible results 

Sure, totally agree, I am looking further into it.

Oh It actually doesnt solve the original problem. This test here produces sensible results https://github.com/SpikeInterface/spikeinterface/issues/3383#issuecomment-2340197858 but plot_unit_waveforms is off 

ok I updated the function get_channel_location also in sorting analyzer 

This was solved yes!