Hi Ronan and COBRA team,
We are currently still optimizing our protocols for running XomicsToModel. One of the outstanding issues with our output models generated from the base Recon3D reconstruction is that they retain metabolite leaks, which contaminate downstream analyses. We have considered beginning with Recon3DModel (fully leakless), but the concern is that there are reactions/metabolites/genes missing from the model that are present in the reconstruction.
To ameliorate this, we have attempted to (1) remove leaks from our contextualized model post-hoc and (2) add back reactions to Recon3DModel while avoiding re-introducing leaks.
What is the recommended starting point for XomicsToModel? If the full Recon3D reconstruction the recommended starting point, do you have suggestions on how to handle metabolite leaks in the Xomics output models? Do you have additional context on the content diff between the full Recon3D reconstruction and the Recon3DModel?
Thank you in advance for your time and guidance!
Best,
Dana