File I/O (spike_sort.io)

Functions for reading and writing datafiles.

Read/Write Filters (spike_sort.io.filters)

Filters are basic backends for read/write operations. They offer following methods:

  • read_spt() – read event times (such as spike times)
  • read_sp() – read raw spike waveforms
  • write_spt() – write spike times
  • write_sp() – write raw spike waveforms

The read_* methods take usually one argument (datapath), but it is not required. The write_* methods take datapath and the data to be written.

If you want to read/write you custom data format, it is enough that you implement a class with these functions.

The following filters are implemented:

filters.BakerlabFilter(conf_file) Filter for custom binary data structure.
filters.PyTablesFilter(fname[, mode]) Read/Write HDF5

Export tools (spike_sort.io.export)

These tolls take one of the io.filters as an argument and export data to the file using write_spt or write_sp methods.

export.export_cells(io_filter, node_templ, …) Export discriminated spike times of all cells to a file.

Reference

Filters are basic backends for read/write operations. They offer following methods:

  • read_spt – read event times (such as spike times)
  • write_spt – write spike times
  • read_sp – read raw spike waveforms
  • write_sp – write raw spike waveforms
class spike_sort.io.filters.BakerlabFilter(conf_file)

Filter for custom binary data structure.

The binary data consists of independent files for each contact in each electrode written as 16-bit ints.

The paths to the datafiles are defined in .inf file that is JSON-compatible and contains at least the following attributes:

fspike : str
path to raw recordings relative to dirname
cell : str
path to spike times (with resolution 20 us) relative to dirname
n_contacts : int
number of contacts per electrode
dirname : str
path to the data
FS : int
spike sampling frequency

Each of the paths can include any of the following Python formatting placeholders:

  • {subject} – subeject name
  • {cell_id} – cell id
  • {ses_id} – session id
  • {el_id} – electrode id

These will be substituted by data extracted from datapath paramaeters of the reading and writing methods.

Parameters:

conf_file : str

path to the configuration file

Methods

close()
read_sp(dataset[, memmap]) Reads raw spike waveform from file in bakerlab format
read_spt(dataset) Returns spike times in miliseconds:
write_sp(sp_dict, dataset) Write raw spike waveform to a file in bakerlab format
write_spt(spt_dict, dataset[, overwrite]) Write spike times to a binary file.
read_sp(dataset, memmap=None)

Reads raw spike waveform from file in bakerlab format

Parameters:

dataset : str

dataset path (in format /{subject}/session{ses_id}/el{el_id})

memmap : {‘numpy’, ‘tables’, None}, optional

if True use memory mapped arrays to save some memory (defaults to no memmory-mapping)
read_spt(dataset)

Returns spike times in miliseconds:

Parameters:

dataset : str

dataset path in format /{subject}/session{ses_id}/el{el_id}/cell{cell_id}
write_sp(sp_dict, dataset)

Write raw spike waveform to a file in bakerlab format

Parameters:

sp_dict : dict

spike waveform dict

dataset : str

dataset path

See also

read_sp

write_spt(spt_dict, dataset, overwrite=False)

Write spike times to a binary file.

Parameters:

dataset : str

dataset path

See also

read_spt

class spike_sort.io.filters.PyTablesFilter(fname, mode='a')

Read/Write HDF5

HDF5 is a hierarchical datafile – data is organised in a tree. The standard layout is:

/{SubjectName}/
/{SubjectName}/{SessionName}/{ElectrodeID}/
/{SubjectName}/{SessionName}/{ElectrodeID}/stim: stimulus time
/{SubjectName}/{SessionName}/{ElectrodeID}/raw: spike waveforms
/{SubjectName}/{SessionName}/{ElectrodeID}/{CellID}: spike waveforms
/{SubjectName}/{SessionName}/{ElectrodeID}/{CellID}/spt: spike
times

where curly brackets {} denote a group.

This layout may be adjusted by changing paths

Methods

close()
close_all()
read_sp(dataset) Read continous waveforms (EEG, LFG, spike waveform)
read_spt(dataset) Read event times (such as spike or stimulus times).
write_sp(sp_dict, dataset[, overwrite]) Write signal
write_spt(spt_dict, dataset[, overwrite]) Write spike times
read_sp(dataset)

Read continous waveforms (EEG, LFG, spike waveform)

Parameters:

dataset : str

path pointing to cell node
read_spt(dataset)

Read event times (such as spike or stimulus times).

Parameters:

dataset : str

path pointing to cell node
write_sp(sp_dict, dataset, overwrite=False)

Write signal

write_spt(spt_dict, dataset, overwrite=False)

Write spike times

spike_sort.io.export.export_cells(io_filter, node_templ, spike_times, overwrite=False)

Export discriminated spike times of all cells to a file.

Parameters:

io_filter : object,

read/write filter object (see spike_sort.io.filters)

node_templ : string

string identifing the dataset name. It will be passed to IOFilters.write_spt method. It can contain the {cell_id} placeholder that will be substituted by cell identifier.

spt_dict : dict

dictionary in which keys are the cell IDs and values are spike times structures