rutishauserlab / recogmem-release-nwb Goto Github PK

I'm trying to use the Matlab code to load the data, but I get an error which seems to be related to datenum8601.m (dnum is an empty array):

Index exceeds the number of array elements (0).

Error in types.util.checkDtype (line 142)
datetime(dnum(1), 'TimeZone', tz, 'ConvertFrom', 'datenum');

Error in types.core.NWBFile/validate_session_start_time (line 470)
val = types.util.checkDtype('session_start_time', 'isodatetime', val);

Error in types.core.NWBFile/set.session_start_time (line 321)
obj.session_start_time = obj.validate_session_start_time(val);

Error in types.core.NWBFile (line 163)
obj.session_start_time = p.Results.session_start_time;

Error in NwbFile (line 15)
obj = [email protected](varargin{:});

Error in io.parseGroup (line 69)
parsed = NwbFile(kwargs{:});

Error in nwbRead (line 31)
nwb = io.parseGroup(filename, h5info(filename), Blacklist);

Error in NWBloadDataOfBlock_release (line 29)
nwb = nwbRead([basepath, filesep, NWBsession.filename]);

Error in NWBprepareData_forSession_demo (line 11)
RTRecog_raw, RTRecog_vsQ] = NWBloadDataOfBlock_release( NOsession_toUse, ind, 1,basepath ); %1 mode recog

Error in NWBneural_loopOverSessions_release (line 33)
[paramsForRun, basedirData_forSession,basedirEvents_forSession,brainArea] = NWBprepareData_forSession_demo( basepath,
NOsessionRecog, NOind, paramsIn );

Error in NWBneural_main (line 94)
[totStats,cellStatsAll] = NWBneural_loopOverSessions_release(allSessionsToUse, NWBsessions, basepathData,analysisMode,
paramsInPreset );

So far I'm just trying to load the data from a single session (session 5, as indicated in NWBneural_main.m).

I've tried with a couple of different versions of matlab (R2021a and R2021b) running in Ubuntu.

Thank you for any help, best

Could you test the code with the latest releases for pynwb and hdmf? I think it should work now

There's no need to pip install pynwb separately, since it is listed in the requirements in the setup.py file. In fact, installing separately may cause issues, since you are locking to a specific version of pynwb

For the dataset that is published on DANDI here, it looks like the spike time are in milliseconds, but they are supposed to be in seconds. This is going to cause problems with any NWB tools that try to run any analysis or visualization on these files.

Sessions that are from the same subject have different subject IDs.nwb.general.subject.subject_id for P9HMH_NOID5.nwb is "P9HMH_NOID5" but I think it should be "9", "H09" or "P9HMH". Sessions that have subject_ids that are the same in the first part also have the same subject attributes, so it seems pretty clear that these are actually from the same patient. subject_id is our way of indicating that. Right now, it appears that these sessions came from different patients because they have different subject_id values.