Question 1

Should a new reconstruction be built or an existing one adopted?

Accepted Answer

Building from scratch is justified when no model exists for the organism, or when the reconstruction process itself is the contribution. Where a curated model already exists, adapting it and documenting every change usually yields more biology per unit of effort. Either way, the version used and all modifications should be recorded explicitly.

Question 2

Why does the biomass equation attract so much scrutiny?

Accepted Answer

Because it defines what the model treats as growth, and therefore shapes almost every flux prediction. Many reconstructions inherit a biomass composition measured in a different organism entirely. Measuring macromolecular fractions for the strain in question, or at minimum testing how sensitive conclusions are to that composition, repays the effort.

Question 3

How should omics data be integrated into a model?

Accepted Answer

Carefully, since methods differ in how they translate expression into flux bounds and can yield conflicting condition-specific models from identical data. The chosen algorithm, its thresholds and the resulting model size should all be reported. Comparing at least two integration methods shows whether a conclusion depends on that choice.

Question 4

What counts as adequate model validation?

Accepted Answer

Comparison against data the model was not fitted to. Growth or no-growth predictions across defined media, gene essentiality against a knockout screen, and predicted against measured uptake and secretion rates are all standard. Reporting where predictions fail, and analysing why, carries more weight than a headline accuracy figure.

Question 5

Do strain design predictions need experimental testing?

Accepted Answer

Not universally, but a purely computational thesis should state that limit plainly and compensate with stronger validation of the model itself. Where testing is planned, constructing and characterising even a small number of predicted strains adds considerable weight, so collaboration and timeline should be arranged early.

Question 6

Which quality standards apply to a reconstruction?

Accepted Answer

Community test suites check mass and charge balance, stoichiometric consistency, annotation coverage, identifier standardisation, and the absence of energy-generating cycles that produce something from nothing. Running these and reporting the scores has become a normal expectation, and unresolved failures should be discussed rather than omitted.

Question 7

How much programming ability is required?

Accepted Answer

Enough for reproducible analysis. Established toolboxes handle simulation, but curation, data integration and custom design algorithms all involve scripting, and reviewers increasingly expect versioned code and models deposited alongside the thesis. Comfort with a scripting language and version control is effectively part of the methodology.

Biotechnology

Biotechnology

Research Frontiers in Metabolic Reconstruction Network Analysis