This comes from a discussion with @rly.
Currently, when loading LF, we load the probe from the AP meta:
|
if "nidq" not in self.stream_id and not load_sync_channel: |
|
signals_info_dict = {e["stream_name"]: e for e in self.neo_reader.signals_info_list} |
|
meta_filename = signals_info_dict[self.stream_id]["meta_file"] |
|
# Load probe geometry if available |
|
if "lf" in self.stream_id: |
|
meta_filename = meta_filename.replace(".lf", ".ap") |
|
probe = probeinterface.read_spikeglx(meta_filename) |
|
|
The reason is that the shank map is only available in the AP meta, and that is needed to load the probe:
#596 (comment)
I propose that if the AP meta is not present for any reason, we load the data without the probe so users can at least access it. I think this might prevent loading the inter_sample_shift property?
But maybe with all the new metadata changes in Neuropixels, we can retrieve the probe information and the inter-sample shift in a simpler way.
Thoughts?
This comes from a discussion with @rly.
Currently, when loading LF, we load the probe from the AP meta:
spikeinterface/src/spikeinterface/extractors/neoextractors/spikeglx.py
Lines 76 to 83 in e45a8be
The reason is that the shank map is only available in the AP meta, and that is needed to load the probe:
#596 (comment)
I propose that if the AP meta is not present for any reason, we load the data without the probe so users can at least access it. I think this might prevent loading the inter_sample_shift property?
But maybe with all the new metadata changes in Neuropixels, we can retrieve the probe information and the inter-sample shift in a simpler way.
Thoughts?