overlapping event latency in epoch2continuous

[anomalosepia]

Description

I don't think that this has been raised yet. If you are converting epoched to continuous and there are events with the same latency in the EEG.event.latency, you lose all but the first event (in the order in the latency struct) during the conversion. This is silent and doesn't raise errors or warnings.

]

Steps to Reproduce

1. Get some continuous data with event structure that looks something like below. This is from EGI netstation so the data can get kind of mangled along the way.

duration	type	latency	epoch	urevent
600	Scrt'	99601	167	602
1	rscrt'	99685	167	603
0	'TRSP'	100193	167	604
0	'TRSP'	100201	168	605
600	'Fix'	100201	168	606
0	'bgin'	100274	168	607
0	'fix+'	100284	168	608

2. run this on your data EEG = epoch2continuous(EEG)

3. the resulting structure will not have the 'Fix' event at 100201 because line 27 in epoch2continuous only takes the first event at any given latency

[xlat, indx] = unique_bc2([EEG.event.latency], 'first');

in my case I was able to change that to

[xlat, indx] = unique_bc2([EEG.event.latency], 'last');

but that's only for the very narrow case of the data that I have here.

Expected behavior:

I expected all the events to carry over from the epoched data to the continuous, with boundary event. data

Actual behavior:

Only the first event comes over.

Versions

OS version	mac os x catalina
Matlab version	2018b
EEGLAB version	2019.1
ERPLAB version	7.0

[andrewxstewart]

Thanks for reporting this bug with epoch2continuous.

I'll take a look at this.

[anomalosepia]

Thank you! I just removed the requirement for unique latencies and it seems to be sorted out. Still checking though. A

Sent from my iPhone

On Feb 14, 2020, at 5:42 PM, Andrew X Stewart notifications@github.com wrote:

 Thanks for reporting this bug with epoch2continuous.

I'll take a look at this.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.