Added OpenEphys to KWD conversion #44
Conversation
There was a problem hiding this comment.
if you're interested in getting klusta-pipeline python3 compatible I could probably get a PR for that going. Want to take a look at my python3 PRs for ephys and behav analysis?
We probably want to take some of the good parts of make_kwd and mix them in. Also, the way the repo is currently organized is a lot of the functions are in dataio and utils which might be useful for you. Both for putting your functions in there and for re using those functions.
| return list(set(recordings)) | ||
|
|
||
| def check_n_samples_consistency(continuous_files): | ||
| n_samples_list = [] |
There was a problem hiding this comment.
why not make this a set to begin with?
| def get_openephys_channel_number(continuous_file): | ||
| cf_filename = os.path.split(continuous_file)[1] | ||
| cf_filename = cf_filename.replace('.continuous', '') | ||
| if 'CH' in cf_filename: |
There was a problem hiding this comment.
if you're going to have these kinds of magic numbers can you do some asserts to make sure you're getting the data you expected.
also maybe elifs and a else case that raises an error?
| kwd_file.create_dataset("/recordings/{}/application_data/channel_sample_rates".format(recording_number), data=channel_sample_rates) | ||
| kwd_file.create_dataset("/recordings/{}/application_data/channel_bit_volts".format(recording_number), data=channel_bit_volts) | ||
| kwd_file.create_dataset("/recordings/{}/application_data/timestamps".format(recording_number), data=np.array(channel_timestamps)) | ||
|
|
There was a problem hiding this comment.
Can we set it up to have a an if name == main so we can run it as a script
| channel_sample_rates[channel] = int(cf_header['sampleRate']) | ||
| channel_bit_volts[channel] = float(cf_header['bitVolts']) | ||
|
|
||
| (n_records, n_oe_file_samples) = calculate_openephys_continuous_sizes(oe_file) |
There was a problem hiding this comment.
you don't need the parenthesis around (n_records, n_oe_file_samples)
| oe_file.seek(recording_offset) | ||
| record_recording_number = np.fromfile(oe_file, np.dtype('>u2'), 1)[0] | ||
| recordings.append(record_recording_number) | ||
| return (recordings, timestamps) |
There was a problem hiding this comment.
you don't need parenthesis here either
| recordings = [] | ||
| timestamps = [] | ||
| (n_records, n_samples) = calculate_openephys_continuous_sizes(oe_file) | ||
| for record in range(n_records): |
There was a problem hiding this comment.
this record/record offset code is a bit of a repeated motif throughout several functions, it might be clearer to say
for record_offset in gen_record_offsets(n_records):
timestamps.append(seek_and_read(oe_file, record_offset + RECORD_TIMESTAMP_OFFSET, '<i8'))
recordings.append(seek_and_read(oe_file, record_offset + RECORD_RECORDING_OFFSET, '>u2')[0])
but that's just my style
| def get_n_channels(continuous_files): | ||
| return len(continuous_files) | ||
|
|
||
| def initialize_kwd_file(data_dir, experiment_name): |
There was a problem hiding this comment.
have you looked at from klusta_pipeline.dataio import save_info, save_recording, save_chanlist, save_probe, save_parameters
there might be a bunch of code you can reuse in there
Also, are we going to have averaging options here? as in klusta_pipeline.utils? options to ignore dead channels?
|
|
||
| # For each record, save the samples in the kwd recording dataset | ||
| for record_ind, record in enumerate(recording_records): | ||
| record_data_list = load_openephys_continuous_record(oe_file, record) |
There was a problem hiding this comment.
maybe change this to
_, _, _. record_data = load_openephys_continuous_record(oe_file, record) or
record_data = load_openephys_continuous_record(oe_file, record)[3]
What metadata needs to get added to the KWD file along with the raw waveforms?