Closed jonrkarr closed 3 years ago
I think errors aren't raised earlier because the HDF5 library has some ability to work with pandas.
This is the same as #82 I posted in July.
Hi Jonathan, I'm trying to locate some omex archive for my testing, most of what I needed I found in https://github.com/biosimulators/Biosimulators_test_suite/tree/dev/examples/sbml-core, but I'm unable to find the following: results_report.SimulatorGeneratesReportsOfSimulationResults sedml.SimulatorProducesLinear2DPlots sedml.SimulatorProducesLogarithmic2DPlots sedml.SimulatorProducesMultiplePlots Can you please help? Thanks!
The test cases are generated dynamically from the COMBINE archives in https://github.com/biosimulators/Biosimulators_test_suite/tree/dev/examples/sbml-core.
This repo provides a command-line program which runs the tests which generate these COMBINE archives dynamically. The command-line program can be used to run individual tests, as well as export the generated archives for inspection. There's information about how to run the command-line program at https://docs.biosimulators.org/Biosimulators_test_suite/tutorial.html#saving-the-synthetic-combine-archives-generated-by-the-test-cases
These command-line options may be helpful:
--test-case sedml.SimulatorSupportsModelAttributeChanges
--synthetic-archives-dir /path/to/export/generated-combine-archives/
--work-dir /path/to/save/outputs-of-combine-archives
--debug
--cli /path/to/VCell-CLI
I'm adding a check for good NumPy data types in biosimulators-utils 0.1.121.
I'm adding a check for good NumPy data types in biosimulators-utils 0.1.121.
thanks! I merged your PR to update the requirements
The change I made doesn't fix the problem. It just adds an earlier and more informative error message.
of course :)
Fixed in vcell git commit a9fa4a0
This is the reason for the failure in biosimulators/Biosimulators#403
Here VCell is converting the data to a numpy array and setting the dtype to
float64
. https://github.com/virtualcell/vcell/blob/9afe44f074116dcf9c8693691ae1d83944a16356/vcell-cli-utils/vcell_cli_utils/cli.py#L217In other places, the VCell code isn't doing this conversion. The dtype needs to consistently be set to a numerical type (e.g.,
float64
for results of continuous simulation,int64
for discrete simulations).Specifically, what needs to be done is to convert each pandas data frame (results of
pandas.read_csv
) to a numpy.array (i.e. callpandas.DataFrame.to_numpy()
). I don't think thedtype
argument needs to be provided. The default dtype should be fine.