Description of the issue:
The merging of HMR and Recon3D results in a model that is imbalanced in both energy and mass in many areas. Extensive automatic and manual curation checks are able to trace these imbalances to specific "problem" reactions which are causing the imbalance.
Some of these "problem" reactions are clear errors, and should simply be removed from the model, whereas others are the result of merging two models with different treatment/assumptions regarding objects such as pools (e.g., lipid pools). When these assumptions/formulations are inconsistent, it results in inconsistent mass/energy balances.
For these latter cases where the problematic reactions are technically correct in many ways, but not sufficiently specific/curated, I propose that they remain in the model, but with their upper and lower bounds constrained to zero. In this way, we preserve the information of the reaction(s) (including potentially important reaction-gene associations), but prevent their ability to disrupt model balances.
Expected feature/value/output:
A mass- and energy-balanced model that cannot generate or destroy mass or energy.
Current feature/value/output:
Currently, the humanGEM is imbalanced, and is able to generate infinite mass (e.g., glucose) and energy (e.g, ATP) from nothing.
I hereby confirm that I have:
Description of the issue:
The merging of HMR and Recon3D results in a model that is imbalanced in both energy and mass in many areas. Extensive automatic and manual curation checks are able to trace these imbalances to specific "problem" reactions which are causing the imbalance.
Some of these "problem" reactions are clear errors, and should simply be removed from the model, whereas others are the result of merging two models with different treatment/assumptions regarding objects such as pools (e.g., lipid pools). When these assumptions/formulations are inconsistent, it results in inconsistent mass/energy balances.
For these latter cases where the problematic reactions are technically correct in many ways, but not sufficiently specific/curated, I propose that they remain in the model, but with their upper and lower bounds constrained to zero. In this way, we preserve the information of the reaction(s) (including potentially important reaction-gene associations), but prevent their ability to disrupt model balances.
Expected feature/value/output:
A mass- and energy-balanced model that cannot generate or destroy mass or energy.
Current feature/value/output:
Currently, the humanGEM is imbalanced, and is able to generate infinite mass (e.g., glucose) and energy (e.g, ATP) from nothing.
I hereby confirm that I have: