Skip to content


Repository files navigation

physicellcdataloader logo & title


physicelldataloader (pcdl) provides a platform independent, python3 based, pip installable interface to load output, generated with the PhysiCell agent based modeling framework, into python3.

pcdl was forked from the original PhysiCell-Tools python-loader implementation.

The pcdl python3 library maintains three branches branches:

  • Branch version 1 is the original PhysiCell-Tools/python-loader code.
  • Branch version 2 will be strictly compatible with the original PhysiCell-Tools/python-loader code, although pip installable.
  • Branch version 3 might break with old habits, although tries to be as much downward compatible as possible. The aim of the v3 branch is to get a very lean and agile python3 physicell output interface, for the ones coming from the python world.


HowTo Guide:


Reference Manual:


To be developed.

About Documentation:

Within the pcdl library, we tried to stick to the documentation policy lined out by Daniele Procida in his "what nobody tells you about documentation" talk at PyCon 2017 in Portland, Oregon.


  • original PhysiCell-Tools python-loader implementation: Patrick Wall, Randy Heiland, Paul Macklin
  • fork pcdl implementation: Elmar Bucher
  • fork pcdl co-programmer and tester: Heber Rocha, Marshal Gress


  author    = {Bucher, Elmar and Wall, Patrick and Gress, Marshal and Rocha, Heber and Heiland, Randy and Macklin, Paul},
  title     = {elmbeech/physicelldataloader: pcdl platform-independent, pip-installable interface to load PhysiCell agent-based modeling framework output into python3.},
  year      = {2023},
  copyright = {Open Access},
  doi       = {10.5281/ZENODO.8176399},
  publisher = {Zenodo},

Road Map:

Release Notes:

  • version 3.2.12 (2023-08-12): elmbeech/physicelldatalodader

    • add pcdl_repl_programming.ipynb : Jupyter notebook to give an idea about how to work with pcdl in a Python3 REPL.
    • add github continuous integration, all supported python versions, all supported operating systems.
  • version 3.2.11 (2023-08-08): elmbeech/physicelldatalodader

    • pip install pcdl: will only install the bare minimum library dependencies.
    • pip install pcdl[data]: will install the minimum dependencies plus the dependencies to download the test dataset.
    • pip install pcdl[scverse]: will install the minimum dependencies plus the dependencies needed to generate anndata object.
    • pip install pcdl[all]: will always install all dependencies.
    • new TimeSeries get_annmcds_list function, which, points to the self.l_annmcds object.
    • new pyMCDS get_scatter function split off from pyMCDSts make_imgcell.
    • pyMCDSts make_imgcell and make_imgsubs bug fixes.
    • TimeStep and TimeSeries get_anndata evolution.
  • version 3.2.10 (2023-07-24): elmbeech/physicelldataloader

    • rename pyMCDSts get_cell_df_columns_states to get_cell_df_states for conciseness.
    • rename pyMCDSts get_conc_df_columns_states to get_conc_df_states for conciseness.
  • version 3.2.9 (2023-07-23): elmbeech/physicelldataloader

    • new class TimeStep can do everything pyMCDS can do and more.
    • new class TimeSeries can do everything pyMCDSts can do and more.
    • new TimeStep get_anndata function to transform physicell output into AnnData objects.
    • new TimeSeries get_anndata function to transform physicell output into AnnData objects.
    • internal pyAnnData scaler function.
    • internal pyAnnData _anndextract function.
    • pyMCDS __init__ seetingxml parameter changed from boolean to string to accept other PhysiCell_settings.xml filenames than the default.
    • pyMCDS get_cell_df drop and keep parameters to declare a set of columns to be dropped or kept.
    • pyMCDS get_concentration_df drop and keep parameters to declare a set of columns to be dropped or kept.
    • new pyMCDS get_conc_df shorthand for get_concentration_df.
    • pyMCDSts get_cell_minstate_col reimplementation as get_cell_df_columns_states function.
    • pyMCDSts get_concentartion_minstate_col reimplementation as get_conc_df_columns_states function.
    • new pyMCDSts get_mcds_list function which, points to the self.l_mcds object.
  • version 3.2.8 (2023-06-21): elmbeech/physicelldataloader

    • pyMCDS get_concentration_df states parameter to filter out non-informative variables.
    • pyMCDS get_cell_df states parameter to filter out non-informative variables.
    • pyMCDSts __init__ load parameter to specify if the whole time series data straight at object initialization should be loaded.
    • new pyMCDSts get_cell_minstate_col function to scan the whole time series for informative features.
    • new pyMCDSts get_concentartion_minstate_col function to scan the whole time series for informative features.
  • version 3.2.7 (2023-06-20): elmbeech/physicelldataloader

    • pyMCDS and pyMCDSts __init__ custom_type parameter to specify other custom_data variable types (int, bool, str) then the generic float.
  • version 3.2.5 (2023-06-19): elmbeech/physicelldataloader

    • pyMCDS resolve incompatibility with earlier PhysiCell and MultiCellDS versions.
  • version 3.2.4 (2023-06-17): elmbeech/physicelldataloader

    • pyMCDS __init__ seetingxml parameter for cases where in the output folder no PhysiCell_settings.xml can be found.
    • pyMCDSts mcdsts.make_imgcell extrema parameter replaced by z_axis parameter to account for numerical and categorical variable types.
  • version 3.2.2 (2023-06-16): elmbeech/physicelldataloader

    • pyMCDS mcds.get_cell_df sets distinct boolean, categorical, integer number, and real number variable types. categorical number codes are translated. for all spatial variables, the vector length value is calculated and added automatically.
    • new pyMCDS mcds.get_celltype_dict function.
    • new pyMCDS mcds.get_substrate_dict function.
    • pyMCDSts mcdsts.make_imgcell and mcdsts.make_imgsubs functions improved.
  • version 3.2.1 (2023-06-12): elmbeech/physicelldataloader

    • pypa odyssey is coming to an end.
    • change build system from setuptools to hatching.
    • change the library name from pcDataLoader to pcdl.
    • to make the library installation more lightweight, test data was excluded from the basic installation. given the computer is connected to the internet, test data can easily be installed and removed with the pcdl.install_data() and pcdl.uninstall_data() functions now.
  • version 3.0.7 (2023-06-08): elmbeech/physicelldataloader

    • pyMCDSts: replaces the svg dependent mcdsts.make_jpeg, mcdsts.make_png, and mcdsts.make_tiff with mcdsts.make_imgcell and mcdsts.make_imgsubs which generate images straight out of the loaded data. the mcdsts.make_gif and mcdsts.make_movie functions were adjusted accordingly.
    • pyMCDSts: mcdsts.read_mcds loads now automatically all mcds snapshots, if no xmlfile_list is provided (default).
  • version 3.0.6 (2023-04-29): elmbeech/physicelldataloader

    • pyMCDS _read_xml is now able to load time steps with zero cells.
    • pyMCDS mcds.get_contour can handle more input parameters.
  • version 3.0.5 (2023-02-26): elmbeech/physicelldataloader pyMCDS mcds.get_contour plots span now the whole domain and not only to the border voxel centers.

  • version 3.0.4 (2023-02-21): elmbeech/physicelldataloader pyMCDS mcds.get_contour function, to easily generate for substrates matplotlib contourf and contour plots because they do not exist as pandas plots.

  • version 3.0.3 (2023-02-19): elmbeech/physicelldataloader branch 3 has no longer anndata and as such hdf5 dependency.

  • version 3.0.2 (2023-01-06): elmbeech/physicelldataloader bugfix installing package data.

  • version 3.0.0 (2023-01-06): elmbeech/physicelldataloader

    • pyMCDS parameter xml_file can now handle path/file.xml (unix) or path\file.xml (dos) input, as long output_path is the default.
    • pyMCDS has a new additional boolean microenv parameter, to specify if the microenvironment (substrates) should be read (for completeness) or not (for speed increase and less memory usage).
    • pyMCDS has a new additional boolean graph parameter, to specify if the attached and neighbor graph should be read.
    • pyMCDS has a new additional boolean verbose parameter, to specify if there should be text output while processing.
    • pyMCDS mcds.get_2D_mesh was renamed to mcds.get_mesh_2D for consistency.
    • pyMCDS mcds.get_linear_voxels was renamed to mcds.get_mesh_coordinate for consistency.
    • pyMCDS mcds.get_containing_voxel_ijk was renamed to mcds.get_voxel_ijk for briefness.
    • pyMCDS mcds.get_voxel_spacing returns now 3 specific values, one for x, y, and z, instead of 1 general value.
    • pyMCDS mcds.get_concentrations was renamed to mcds.get_concentration for consistency
    • pyMCDS mcds.get_concentrations_at was renamed to mcds.get_concentration_at for consistency
    • pyMCDS mcds.get_concentration_at if z_slice is not a mesh center value, the function will by default adjust to nearest and no longer break.
    • pyMCDS mcds.get_cell_variables and mcds.get_substrate_names return now a strictly alphabetically ordered list.
    • pyMCDS mcds.get_cell_df returns now a pandas dataframe with the cell IDs as the index and not as a column. additionally, this dataframe contains now voxel, mesh_center, substrate parameter, substrate concentration, and cell density information too.
    • new pyMCDS mcds.get_concentration_df function.
    • new pyMCDS mcds.get_substrate_df function.
    • new pyMCDS mcds.get_unit_se function.
    • new pyMCDS mcds.get_multicellds_version function.
    • new pyMCDS mcds.get_physicell_version function.
    • new pyMCDS mcds.get_runtime function.
    • new pyMCDS mcds.get_timestamp function.
    • new pyMCDS mcds.get_voxel_ijk_range function.
    • new pyMCDS mcds.get_voxel_ijk_axis function.
    • new pyMCDS mcds.get_voxel_spacing function.
    • new pyMCDS mcds.get_voxel_volume function.
    • new pyMCDS mcds.get_mesh_mnp_range function.
    • new pyMCDS mcds.get_mesh_mnp_axis function.
    • new pyMCDS mcds.get_xyz_range function.
    • new pyMCDS mcds.is_in_mesh function.
    • new pyMCDS mcds.get_attached_graph_dict function.
    • new pyMCDS mcds.get_neigbor_graph_dict function.
    • class pyMCDS_timeseries was renamed to pyMCDSts and completely rewritten.
    • new pyMCDSts get_xmlfile_list function.
    • new pyMCDSts read_mcds function.
    • new pyMCDSts make_jpeg function.
    • new pyMCDSts make_png function.
    • new pyMCDSts make_tiff function.
    • new pyMCDSts make_gif function.
    • new pyMCDSts make_movie function.
    • all plotting functions were removed because pcdl only focus on making the raw data in python easy accessible for in-depth analysis.
    • cell position coordinates are now constantly labeled as x,y,z, mesh center coordinates as m,n,p, and voxel coordinates as i,j,k.
    • the underling mcds object data dictionary structure has changed.
    • pytest unit tests exist now for all pyMCDS and pyMCDSts functions.
  • version 2.0.3 (2023-06-16): elmbeech/physicelldataloader pypa odyssey is coming to an end.

  • version 2.0.2 (2023-01-06): elmbeech/physicelldataloader reset patch voxel spacing bugfix, so that branch2 is full compatible with branch1 again. use branch3 for a bugfixed version!

  • version 2.0.1 (2022-11-08): elmbeech/physicelldataloader beta release patch voxel spacing bugfix.

  • version 2.0.0 (2022-08-30): elmbeech/physicelldataloader pip installable release, derived from and compatible with PhysiCell-Tools/python-loader release 1.1.0 (2022-07-20).

  • version 1.1.1 (2022-07-01): elmbeech/physicelldataloader deprecated np.float replaced with np.float64.

  • version 1.1.0 (2022-05-09): Physicell-Tools/python-loader release compatible with pre-v1.10.x of PhysiCell.

  • version 1.0.1 (2020-01-25): Physicell-Tools/python-loader time-series related bug fix.

  • version 1.0.0 (2019-09-28): Physicell-Tools/python-loader first public release!