mne-tools / mne-incubator Goto Github PK

here is a new package by @TomDLT on CFC

https://pactools.github.io/index.html

cc @choldgraf @SherazKhan

I am tempted to remove the CFC code from sandbox and tell people to use this package

any opinion?

@choldgraf could you take a look at the cfc code? MNE-Python has deprecated the ctw_morlet function, so now mne-sandbox doesn't work anymore due to an improper import.

From @teonbrooks and mne-tools/mne-python#2112:

It would be nice if we had some denoising options available for preprocessing data. A few options that I was thinking of tsPCA (Cheveigné, 2007), and CALM (Adachi, 2001), which can be a special case of tsPCA.

there's a package on git (https://github.com/pealco/python-meg-denoise) that contains an implementation, as well as source code (http://www.isr.umd.edu/Labs/CSSL/simonlab/Denoising.html).

use case for those whose lab implement these tools in Matlab, or private software, those don't have license for maxfilter but would like cleaner data.

On all current PRs travis fails for PYTHON=2.7 TEST_LOCATION=install (e.g. https://travis-ci.org/mne-tools/mne-sandbox/jobs/124744671#L3).

I guess there is a problem with creating the symbolic links?

For consistency with our other repositories, the master branch should be renamed to main. I cannot do it myself because I'm lacking permissions.

We should add the faster code to the sandbox at some point. If not the entire pipeline, at least functions like detect_bad_channels, detect_bad_epochs and detect_bad_channels_in_epochs are very useful.

When it was first suggested to me to add some code to this repository, I felt very strongly against it, and I realize part of this objection was due to the repository's name. A "sandbox" can usually mean two things:

• a safe and isolated place where kids can go play without harming themselves or anyone else,
• a safe and isolated space provided by computer software or the operating system where potentially malicious actors are locked in so they cannot destroy the rest of the system

Code that isn't just quite ready for production, but should be worked on with the clear intention of getting integrated upstream sometime in the future falls into neither of these two categories: We're not kids playing, nor are we malicious (at least I hope so!); and we do want to get this stuff production-ready – at least that's what I want to do!

So what the mne-sandbox actually is supposed to be is some sort of a "laboratory" for new ideas, but with a clear path to get merged upstream if the experiments turn out to be fruitful, promising, and "ripe" for general use.

These kinds of repositories are commonly called incubators. See for example:

• Apache Incubator
• Mozilla Builders Incubator
• IBM Open-Source Incubator

just to name a few.

I would therefore like to propose to change the name of this repository to mne-incubator so at a single glance it becomes clear: this is not supposed to be a dusty attic where ideas are just being placed and forgotten, but that we're determined to bring the experiments happening here into production eventually.

I was not very familiar with mne-sandbox, I was trying phase_amplitude_coupling, which was throwing error due to the depreciated cwt_morlet from mne-python in cfc.py.