Releases: resendislab/corda
v0.5.1
Please see https://github.com/resendislab/corda/blob/master/NEWS.md for the full release notes.
v0.4.0
v0.3
This release includes several changes and bug fixes.
Ported to the new cobrapy (0.6.0 alpha)
CORDA now uses the new COBRApy which brings a new solver interface. This gets rid of having to formulate the problem as irreversible reactions as this is the default low-level problem formulation used in COBRApy now. Also brings less convoluted formulations.
Ported to CORDA 2
This includes the changes from the CORDA 2 which brings several improvements. Most importantly CORDA reconstructions are now deterministic and will give the same results for the same input every time. Additionally, the redundancy detection was overhauled and is now more correct and much faster. By default CORDA will now detect all redundant pathways. This gives a significant speed-up for larger models. Reconstruction times for Recon 2 are now in the order of a few minutes.
New helper attributes
The CORDA object now has two new attributes.
CORDA.includednow gives you a dict of booleans telling you whether each reaction was included in the reconstruction or not.CORDA.redundanciesgives you a dict telling you how many additional/redundant pathways are there are to activate the high-confidence reactions. Here 0 means not optimized meaning the reaction was never a target itself, 1 means a unique activation pathway, etc.
Small changes
printfor a CORDA class now only reports statistics for the reversible formulation- the argument
nnow defaults tonumpy.infmeaning find all redundancies
corda 0.2.0
This release brings a more stable handling of the redundancy. In previous versions some non-optimal solutions where possible. This is fixed now which in turn creates even better reconstructions.
Additional fixes where applied to the info function which now show the correct model metrics.
corda 0.1.1
Fixes a bug with the random noise that is used to find redundant pathways. This bug did only affect reconstructions with many more medium confidence reactions than absent ones.