A workspace to play with MEG data
There are a number of dependencies:
ipython-notebook
numpy
scipy
matplotlib
pylab
scikit-learn
pandas
statsmodels
PyMVPA2 (from source [repos are out of date]: https://github.com/PyMVPA/PyMVPA)
You'll need to create a subdirectory named 'MEG_data' and store the relevant data files there for this code to work. Within MEG_data, you'll need:
paired .set and .fdt files containing MEG data
a space-delimited .tab file with word-level predictive features
Once your file structure looks good, if you want to run an interactive notebook session (outdated), you'll need to go to the 'notebooks' subdirectory and type
ipython notebook
or you can do non-notebook-based analyses by going to the 'scripts' directory and typing:
python meg_stats_runner.py
Easy command-line querying of the output from stats_runner can be obtained by typing:
python query_stats.py
Visualization of stats_runner output is available from:
notebooks/Topomap_real
To obtain the V7.tab file with feature annotations:
Begin with the V3.tab file from the MEG HOD corpus.
Obtain the gcgbadwords, gcgparseops, and totsurp files from the authors (or from /home/corpora)
python ../scripts/buildtab.py hod_V3.tab hod.wsj02to21-comparativized-gcg15-1671-4sm.fullberk.parsed.gcgbadwords > hod_V4.tab
python ../scripts/addsentid.py hod_V4.tab > hod_V5.tab
python ../scripts/addparserops.py hod_V5.tab hod.wsj02to21-comparativized-gcg15-1671-4sm.fullberk.parsed.gcgparseops > hod_V6.tab
python ../scripts/addtotsurp.py hod_V6.tab hod.totsurp > hod_V7.tab
Set tokenPropsFile = <.tab filename>