Silvio Waschina

Hi, Thanks for reporting this. I can confirm that there is an inconsistency in the completeness calculations. In principle, the completeness (C) is calculated as: $$ C = \frac{N_{found} +...

Hi @MRMHmdeleeuw Thank you for reporting the issue and providing the model files. I can reproduce the error when using glpk as solver. The issue is most likely due to...

I usually use IBM's CPLEX as solver. This requires the installation of the R-package [cobrarCPLEX](https://github.com/Waschina/cobrarCPLEX/tree/main), and of course IBM ILOG CPLEX itself. But I think I solved the issue with...

The latest cobrar development version now has the bug fix. You can update your cobrar version by running e.g. `remotes::install_github("Waschina/cobrar")` .

Hi @MRMHmdeleeuw , I had a closer look at this again. The special case with the [1KFYEC4J.tgz](https://github.com/user-attachments/files/20959151/1KFYEC4J.tgz) model is that the medium prediction added N2 and no Ammonium to the...

Hi @cdiener as Johannes mentioned, the issue should be fixed now. I tested the `METHANOGENESIS-PW` in three Archaea genomes: 1. _Methanothermobacter thermautotrophicus_ (GCF_000008645.1). Completness: **100%** 2. _Methanobrevibacter smithii_ (GCF_000016525.1). Completness:...

Hi Marco, sorry for this super-late reply. Before, I generated the tables with sume custom script. But now, with the cobrar package and gapseq returning cobrar models, it's easy to...

@jotech , can you help with the two questions?