This includes a list of major changes for each minor version starting from 0.10.0. q2-micom uses SemVer 2.
For information on how to use q2-micom please refer to
the tutorial.
New interaction visualizations that allow tracking metabolite exchanges between taxa.
New visualization shwoing Metabolic Exchange Scores (MES).
Growth rates and taxon exchange plots now allow coloring by custom metadata columns.
minimal-medium now supports more options and outputs global and sample-specific media
by default. We also provide the results associated with the media optimization which can
help to inspect growth rates or required taxon-specific uptakes.
plot-fit got renamed to association and will not use LASSO coefficients anymore as
a proxy for univariate associations. Instead univariate non-parametric tests are run for
each metabolite. This should be more stable and avoid issues with low reproducibility of
coefficients. LASSO model performance is still reported as a global measure of
association. In case you need access to the previous plot-fit function you can simply
install a prior version of q2-micom (<0.14.0) in a Qiime 2 environment.
Some broad updates to the visualizations with updated dependencies and some visual updates to avoid very squished plots.
Now uses the hybrid solver by default which gives better accuracy.
Fixes a bug where the build manifest would always report 100% matching.
Pins the MICOM versions to anticipate the transition from fit -> association.
Fixes the broken prefix trimming.
Now also reports the matched relative abundance in verbose build output.
Fixes issues with the new model databases due to automatic stripping of taxonomic rank prefixes.
Switches the default solver to the new hybrid solver.
Updated the tutorial.
Fixes the CI, no changes to the actual codebase.
Fixes a bug where the build step would crash for datasets with a single sample.
Avoids updating the base Qiime2 scikit-learn which can break taxonomy classifiers.
Fixes for the PYPI upload.
Add some more citations. Now tests against newer Qiime2 versions.
Now always cast sample IDs to strings avoiding issue with numeric IDs.
build now accepts the --p-solver parameter that allows to specify which solver
to use in case one has several installed.
grow now accepts the --p-strategy parameter that allows to specify the strategy
to pick a flux solution.
grow now defaults to using parsimonious FBA (pFBA) to get fluxes as this has more
empirical evidence in returning the biologically relevant flux distributions. In some
instances this can lead to results that are different from what previous versions
have returned. In particular this will may lead to less secreted molecules.
In this case you can pick the previous strategy by using --p-strategy="minimal uptake".
Fixed a bug in build when using the strict option where it would not use the correct
taxonomic rank to merge.
Fixed a bug in the filter functions where the metadata was not correctly transformed to a DataFrame.
Corrected the filter commands in the docs.
Fixed an error where growth results could not be saved to an artifact.
Fixed building of package and artifacts.
Add the actions filter-models and filter-results.
Fixed an issue with the required MICOM version.
This was a maintenance release that added support for newer Qiime2 and MICOM versions.