Step 1B Execution Time Issue

[navya1104]

Hi Tim, thank you for the wonderful tool!

I've been having some issues with step 1B particularly, in the pipeline. I was hoping to know if there are some optimizations possible or any workarounds to have step 1B finish within a reasonable timeframe. Sharing some details below!

The batch I am trying to analyze comprises 5 .raw recordings - each corresponds to a 48 well plate, analyzed with 16 electrodes, and is a 10 mins recording. I am currently using only one threshold and one STTC lag / Node cartography lag value. The batch has 12 groups in analysis. Computer specifications - Mac Pro with 192 GB RAM.

I have found step 1B to become exponentially slower while it is running and is often freezing before completion. For the 5 recordings (240 .mat files), this single sub-step ran for over 2 days without completing fully. Are there any optimizations possible or anything that could be done differently to tackle this running time issue? All other steps are running fine and I have been able to get all outputs skipping step 1B (the spike check plots).

Thanks again for the wonderful work!

[Timothysit]

Hi, thanks for using MEA-NAP! Usually I find step 1A (spike detection) to take the longest (or step 4). So I suspect that step 1B is slow due to perhaps the high activity level you have / something else is not working properly. But first, can you help me check what the running speed is like on our test files? That will help give me an idea about your computer specs.

First, you tick "Show Advanced Settings", then in the Pipeline tab, set Verbose Level to High and tick Time processes. Then click on the Test Pipeline button (note: it will download some test files if it hasn't been downloaded already, just make sure your computer has around 4 GB free space). After it runs, your GUI should look like this:

If you scroll to the top of the command window in MATLAB, it should also tell you how long the spike detection took for each file, mine looks like this:

Elapsed time is 172.950212 seconds.
Elapsed time is 157.695005 seconds.

which suggests that step 1B didn't take long: 425 - 173 - 158 = 94 seconds.
(for reference I ran this on Ubuntu 24.10 with a computer with the AMD Ryzen 7 8745HS chip and 26 GB ram)

Anecdotally I find that MEA-NAP runs faster on my linux machine than my macbook (M1 Pro Chip).

Overall, I suggest you try test pipeline first and we can see whether 1B is unusually slow even for the test files or not. And if it seems to be something to do with the activity level of your spike detected files, then perhaps we can arrange for you to send me one or two test files to test step 1B? Or if you are not comfortable with sharing your data, you can also let me know what the firing rate properties are like so I can simulate some data and test it. Thanks!

[navya1104]

Thanks Tim! This is very helpful, will test it with your suggestions and let you know how it goes this time!