https://github.com/bdsinger/pnicorr
High performance auto-correlation of fMRI data
https://github.com/bdsinger/pnicorr
Last synced: 9 days ago
JSON representation
High performance auto-correlation of fMRI data
- Host: GitHub
- URL: https://github.com/bdsinger/pnicorr
- Owner: bdsinger
- License: mit
- Created: 2014-10-15T23:00:49.000Z (over 11 years ago)
- Default Branch: master
- Last Pushed: 2015-07-13T14:02:49.000Z (almost 11 years ago)
- Last Synced: 2026-06-11T06:27:38.607Z (28 days ago)
- Language: C
- Size: 574 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
pnicorr
=======
High performance auto-correlation of fMRI data
- Intial version supports .1D | .1D.gz format as exported from [afni](http://afni.nimh.nih.gov)'s SUMA program.
- Uses multi-threaded BLAS to do the correlation
- Allows one to provide an upper RAM limit
From the help:
usage: pnicorr file.1D[.gz] [options]
-norm|-nonorm: normalize rows (or don't). default is to normalize.
-mem=MB: memory (MB)
Smaller means more file activity; computing is done
in stages. Default is the same as -mem=4000 (4G)
-iotype=1D|1Dgz|mat: 1D: same as input (SUMA ascii) [default for input]
1Dgz: same as input, gzipped
mat: matlab .mat file [default for output]
Really the input can be any matrix of values, where each row is a time-series to be correlated with all other rows. The format is:
[# comments]
[int int int ] fl.oat fl.oat ...
([] means optional.)
Note that only the metadata columns must be ints and must have no decimal. Data must have a decimal, even if it is just `.0`. This is the quick and dirty way the file is parsed. This is the format saved by the `3dVol2Surf` program from [afni](http://afni.nimh.nih.gov). Other input formats will be supported asap.
Each row is a timeseries with its values separated by whitespace. The rows can be any timeseries data. No spatial information is used (nor available.)