Regarding duplicate reactions in Recon3D

[Juan Sepulveda]

Dear community,

I followed the tutorial_reconToFBAmodel.mlx code to replicate the generation of Recon3D_model but found some issues with how duplicate reactions up to a proton are handled.

1. I noticed that duplicate reactions are kept and removed based on their alphabetical order (i.e. first alphabetical reaction is kept while the second is removed). Is there a better way to decide which duplicate to keep?

2. Some reaction sets are marked as duplicates even though they are not. For example, reaction 1 could be A[c] + h[e] -> h[c] + A[e] while reaction 2 could be A[c] -> A[e]. Should these reactions be treated as duplicates? Do we expect that keeping both of these reactions might create loops?

3. Some reaction sets are marked as duplicates (and they are although they have opposite directions) but they have different GPRs. For example, reaction 1 could be A -> B while reaction 2 could be B -> A. These are technically the same reaction but biologically they are not (as they are catalyzed by different reactions). Do you recommend keeping these duplicate sets? Would you expect that keeping these duplicates would create loops? If so, do you think that these loops could be identified and eliminated by thermokernel downstream?

I appreciate you reading this message and any insight you can provide.

Best,

Juan