Model Differences From wcEcoli

This page documents all the known reasons why a simulation workflow run in vEcoli yields different results from the same workflow run in wcEcoli. Beyond the differences below, it is important to note that vEcoli uses newer versions of dependencies (including Python, NumPy, SciPy) compared to wcEcoli, which may also contribute to differences in simulation results.

This page was written on 2025-10-22 and compares vEcoli commit 931257c with wcEcoli commit b5128d0.

ParCa

• Reformulated RNA degradation Km optimization as a minimization problem and uses Autograd instead of Aesara (not actively maintained). See PR#266 and PR#260.

• Made new gene position data 1-indexed to match other genome sequences and ensure the correct RNA sequence. See commit 3a6da7c from PR#306 and the source code for _build_cistron_data().

• Adjusted select gene transcription and translation efficiencies to better match experimental data and fix some mechanistic issues. See commit 7035fb7 from PR#306, e94e69a from PR#275, and b75278c from PR#243.

Simulation

• Converted ribosome footprint size from nucleotides to amino acids. See commit 8d8fa76 from PR#306.

• Molecules at the same coordinates as replisomes are now removed. See PR#268.

• Superhelical density calculations can now handle removed replisomes. See PR#269.

• Stored mass data in a slightly different Numpy format causing floating point differences. The wcEcoli behavior can be restored by including {"process_configs": {"ecoli-mass-listener": {"match_wcecoli": true}}} in the simulation config JSON. See commit e562507.

• Prevented RNAPs from being initialized outside chromosome domain boundaries. See PR#359.