Reproduce the figures derived from the ODE model described in "Inflammatory Bowel Disease: How Effective Is TNF-α Suppression?" article, published in PLoS One, in 2016, and submit the resulting curated and annotated model to the BioModels database, adhering to their stringent standards.
Since code annotations are present, code-linking has been disabled. Indeed, this warning and error has been reported in Quarto Code Linking documentation
Below, we described remaining steps before final publication to BioModels standards.
- Strange units (such as weeks instead of seconds or days or grams instead of moles)
- May have to check some of the fixed parameters, in particular decay rates of cell populations
- Typos in equation 8 (should be IL-2 both at numerator and denominator)
- General advise: decompose each chemical process into independant annotations (streamline curation, ensure afterwards that the differential equations are properly written, ..)
- There's a differential equation of M0, as being differentiated into M1 and M2. Yet, from the paper, it is assumed to be a fixed influx. Ensure that it is the case in the global variables.
- Done: Able to reproduce the steady-state equations, up-to a normlisation constant (Table 4)
- ToDo: Parameter estimation, after introducing dysregulations, for each of the four endotypes would require to have the omics profiles for all the 58 patients of the cohorts
- Fig2: Sensititvy analyses: without the code and the random seed, may exhbit sensitive discrepancies
- Fig3: Dysregulated mechanisms: infer back the variations of the parameters would require that the omics expression profiles are left avalaible, which is not the case
- Original code scripts, or ODE parameter configurations, to reproduce the results
-
- Which random seed have they used for the PRCC analyses reported in Table 5 and Fig 2?
- Raw transcriptomic profiles to reproduce DGEA summarised results reported in Table 7 (proper parameter estimation reported in Table 8 requires several observations and uncertainty, which is not currently feasible with only one summarised point reported).
- Which parameter estimation optimisation algorithm have they used for Table 8 (between 'DifferentialEvolution', 'SRES', 'EvolutionaryProgram', 'GeneticAlgorithm', 'GeneticAlgorithmSR', 'HookeJeeves', 'LevenbergMarquardt', 'NL2SOL', 'NelderMead', 'ParticleSwarm', 'Praxis', 'RandomSearch', 'ScatterSearch', 'SimulatedAnnealing', 'SteepestDescent' and 'TruncatedNewton') + the hyper-parameters used (number of maximal iterations, tolerance threshold, ...)