opexebo 0.7.2

Collection of python code in Kavli lab.

opexebo

This library provides various analysis tools written by and for the Moser group at the Kavli Institute in Trondheim.

The package contains a mixture of translations of the older MatLab based Behavioural Neurology Toolbox, and other functions written only in Python 3, as part of the transition towards providing open access to research code.

Documentation

Specifications and documentation for opexebo are available on Read the Docs.

Installation

Install with pip:

pip install opexebo

Opexebo has an optional dependency, sep, that is not installed by default. To be able to install it, you require a C++ compiler installed on your system. On Linux, gcc will do the job. On Windows, the the Microsoft Visual C++ Build Tools fulfil the same role (https://www.microsoft.com/en-us/download/details.aspx?id=48159). To force installation of all optional dependencies, append [full] to the install command, for example:

pip install opexebo[full]

Tests

Tests are implemented with pytest. Ensure that the various development tools are installed via

pip install -r dev-requirements.txt

And then run tests with pytest .

History

0.7.0

• Update dependencies for more modern versions (particularly numpy, scipy and pytest)
• Switch supported Python versions for 3.8+

0.6.1

• Fix some docstrings that were not updated to reflect changes in 0.6.0
• Fix the default limit behaviour of accumulate_spatial for bin width specification.

0.6.0

• Rewrite spatial_occupancy and rate_map to remove the walk filter (signature breaking)

  • The individual analysis functions should be single purpose, and the walk-filter breaks this rule.
  • Ideally, the walk-filter should be applied to spikes/tracking data first, and the resulting data then used for further analysis.
  • A dedicated function has been provided to replace this: general.walk_filter

• Rewrite shuffle to better handle edge cases (including 1-spike-only) (signature breaking)

• Added general.walk_filter to provide replace the specific filtering implemented in each of spatial_occupancy and rate_map

• Fix a bug where accumulate_spatial can't handle cases where a single spike is present

• Fix an error in the LFP power spectrum calculation with fft

• Removed support for providing 1D position data as a 2d array with length 1

  • Caused frustrating circular issues in cases where only a single data point was provided.
  • Fairly low value addition, so support removed.

0.5.5 (2021-06-01)

• Publish to PyPi

0.5.2 (2021-04-19)

• Support 1D data in spatial occupancy and ratemap
• Add Opexebo-specific errors
• Add automated testing prior to release
• Add calculate_speed
• Add egocentric_occupancy
• Fixed an edge case error in accumulate_spatial

0.4.3 (2020-02-19)

• Substantial cleanup to accumulate_spatial
  • Function was mathematically correct but confusing, due to the difference between opexebo's standard of (x, y) and NumPy's standard of (y, x)
• Added upsampling
• Added circular_mask
• Mandatory keyword arguments have been made positional
• peak_search updated to handle issues with MaskedArrays containing negative values
• Dcumentation expanded, made Sphinx compatible, and published on ReadTheDocs
• Further work on BorderScore, still experimental though
  • Functional for Rectangular arenas
  • Circular arenas are still WIP, should not be relied upon yet.

0.4.2 (2020-02-05)

• Speedscore modified substantially
  • Adaptive filtering added, allowing the user to specify an upper speed as a bandwidth
  • Adaptive filtering adjusted to behave more sensibly in the case of small bin sizes
  • Speeds are smoothed before correlating
• Spatial cross correlation added
• General bin_number to bin_size code refactored to make future development more consistent
• Assorted unit tests added, including run-once code removed for speed
• Fix errors associated with > where applied to MaskedArrays (which, unlike ndarrays, do not correctly obey the symbol)

0.4.0 (2019-12-10)

• Ratemaps updated to handle walk filter (signature now requires providing animal speeds)
• Angular occupancy coverage calculation corrected
• circular arenas now handled in Spatial Occupancy
• Population Vector Correlation added

0.3.5 (2019-11-18)

• Assorted bugfixes to HDtuning
• Gridscore to handle NaNs more gracefully

0.3.4 (2019-20-25)

• sep made an optional dependency
• Alternative, non sep based code uses skimage
• Attempting to use sep if not installed with raise an error

0.3.2 (2019-09-20)

• Consistent calculating of bin numbers when histogramming

0.3.0 (2019-08-27)

• Third development release
• Improve NaN handling throughout by moving to MaskedArrays
• Fix Angular handling to function in seconds instead of frames
• Remove matplotlib from requirements

0.2.0

• Second development release
  • Implementation of time-map -> rate-map -> (acorr, gridness score)
  • Implementation of rate-map stats, grid stats
  • Implementation of angular-map -> tuning-curve -> head direction score, stats.
  • Non-production-ready implementation of border-score, speed-score.

0.1.0 (2018-08-30)

• First development release
  • Implementation of gridness score, autocorrelogram, place-field detection