Closed nippoo closed 9 years ago
What is the difference with --debug?
Le mardi 6 janvier 2015, nippoo notifications@github.com a écrit :
The --convert-only option allows users to generate .high.kwd, .low.kwd, and .raw.kwd files using klusta without spike detection. A single chunk is detected, but the rest is ignored. This takes much less time than running
the full spike detection algorithm every time.
You can merge this Pull Request by running
git pull https://github.com/klusta-team/spikedetekt2 maxfilter
Or view, comment on, or merge it at:
https://github.com/klusta-team/spikedetekt2/pull/17 Commit Summary
- Added 'convert-only' option to the main script
- Added code for --convert-only flag; detects spikes on first chunk but then only performs format conversion
- More useful message layout to avoid over-verbosity during spike sorting
- Added test for specifying different threshold factors per-channel
File Changes
- M spikedetekt2/core/main.py https://github.com/klusta-team/spikedetekt2/pull/17/files#diff-0 (109)
- M spikedetekt2/core/script.py https://github.com/klusta-team/spikedetekt2/pull/17/files#diff-1 (31)
- M spikedetekt2/processing/tests/test_threshold.py https://github.com/klusta-team/spikedetekt2/pull/17/files#diff-2 (6)
Patch Links:
- https://github.com/klusta-team/spikedetekt2/pull/17.patch
- https://github.com/klusta-team/spikedetekt2/pull/17.diff
— Reply to this email directly or view it on GitHub https://github.com/klusta-team/spikedetekt2/pull/17.
--debug only processes the first chunk. --convert-only converts the whole file, also performing spike detection for the first chunk because otherwise KlustaViewa will refuse to open (this is another bug which we probably don't need to fix before phy).
So this is useful if, for example, you don't want to save a .raw.kwd but then (after clustering) decide you want to use TraceView and look at the spikes or whatever - will convert files in a few minutes rather than a few hours.
Basically this saves a lot of time if you don't want to do spike detection all over again but do want to convert from .dat/.ns5/.bin to .kwd.
OK
BTW @kdharris101 I can't remember what was the last decision regarding .dat/.kwd. How will we store raw data in the next version of the file format (presumable NWB)? With flat binary files or HDF5 files?
The raw data format has not been decided, but will be polymorphic: it could be encoded in a HDF5 file (this is what Allen Inst will probably do) or in native format (what Svoboda lab will do).
From: Cyrille Rossant [mailto:notifications@github.com] Sent: 08 January 2015 12:44 To: klusta-team/spikedetekt2 Cc: Harris, Kenneth Subject: Re: [spikedetekt2] Added --convert-only option and another test (#17)
OK
BTW @kdharris101https://github.com/kdharris101 I can't remember what was the last decision regarding .dat/.kwd. How will we store raw data in the next version of the file format (presumable NWB)? With flat binary files or HDF5 files?
— Reply to this email directly or view it on GitHubhttps://github.com/klusta-team/spikedetekt2/pull/17#issuecomment-69174360.
The --convert-only option allows users to generate .high.kwd, .low.kwd, and .raw.kwd files using klusta without spike detection. A single chunk is detected, but the rest is ignored. This takes much less time than running the full spike detection algorithm every time.