Migration guides

Migration to v3 version

• neurom.view.viewer is deprecated. To get the same results as before, use the replacement:

  import neurom as nm
  # instead of: from neurom import viewer
  from neurom.view import matplotlib_impl, matplotlib_utils
  m = nm.load_morphology('some/data/path/morph_file.asc')
  
  # instead of: viewer.draw(m)
  matplotlib_impl.plot_morph(m)
  
  # instead of: viewer.draw(m, mode='3d')
  matplotlib_impl.plot_morph3d(m)
  
  # instead of: viewer.draw(m, mode='dendrogram')
  matplotlib_impl.plot_dendrogram(m)
  
  # If you used ``output_path`` with any of functions above then the solution is:
  fig, ax = matplotlib_utils.get_figure()
  matplotlib_impl.plot_dendrogram(m, ax)
  matplotlib_utils.plot_style(fig=fig, ax=ax)
  matplotlib_utils.save_plot(fig=fig, output_path=output_path)
  # for other plots like `plot_morph` it is the same, you just need to call `plot_morph` instead
  # of `plot_dendrogram`.
  
  # instead of `plotly.draw`
  from neurom import plotly_impl
  plotly_impl.plot_morph(m)  # for 2d
  plotly_impl.plot_morph3d(m)  # for 3d
  

• breaking features changes:

  • use max_radial_distance instead of max_radial_distances

  • use number_of_segments instead of n_segments

  • use number_of_neurites instead of n_neurites

  • use number_of_sections instead of n_sections

  • use number_of_bifurcations instead of n_bifurcation_points

  • use number_of_forking_points instead of n_forking_points

  • use number_of_leaves instead of number_of_terminations, n_leaves

  • use soma_radius instead of soma_radii

  • use soma_surface_area instead of soma_surface_areas

  • use soma_volume instead of soma_volumes

  • use total_length_per_neurite instead of neurite_lengths

  • use total_volume_per_neurite instead of neurite_volumes

  • use terminal_path_lengths instead of terminal_path_lengths_per_neurite

  • use bifurcation_partitions instead of partition

  • new neurite feature total_area that complements total_area_per_neurite

  • new neurite feature volume_density that complements neurite_volume_density

Migration to v2 version

• Neuron object now extends morphio.Morphology.

• NeuroM does not remove unifurcations on load. Unifurcation is a section with a single child. Such sections are possible in H5 and ASC formats. Now, in order to remove them on your morphology, you would need to call remove_unifurcations() right after the morphology is constructed.

  import neurom as nm
  nrn = nm.load_morphology('some/data/path/morph_file.asc')
  nrn.remove_unifurcations()
  

• Soma is not considered as a section anymore. Soma is skipped when iterating over morphology’s sections. It means that section indexing offset needs to be adjusted by -(number of soma sections) which is usually -1.

• drop benchmarks

• drop neurom.check.structural_checks as MorphIO does not allow to load invalid morphologies, and it does not give access to raw data.

• drop Tree class. Use Section instead as it includes its functionality but if you need Tree separately then copy-paste Tree code from v1 version to your project.

• Section and Neurite class can’t be copied anymore because their underlying MorphIO objects can’t be copied (pickled). Only copying of Neuron is preserved.

• drop FstNeuron. It functionality is included in Neuron class. Use Neuron instead of FstNeuron.

• Validation of morphologies changed.

  The following is not an invalid morphology anymore:

  • 2 point soma

  • non-sequential ids

• script morph_check and morph_stats changed to neurom check and neurom stats

  correspondingly.