Scientific limitations

Interpreting predictions

• A ΔΔG prediction does not predict melting temperature gains, catalytic activity or shelf life. Results state the units, effect direction and assessment source.
• No overall validated product accuracy percentage has been established. Individual model metrics and agreement with literature do not guarantee results for a new protein; disagreement between channels calls for further checks.
• AlphaFold DB supplies precomputed structures. Oligoplan does not run AlphaFold or establish a mutant's structure; pLDDT and PAE describe structural model confidence, rather than experimentally measured stability.
• Donor search is limited by available sources, returned records and filters. A candidate is not declared the best among all existing proteins. Check function, record completeness and numbering.
• Mutation combinations and primers are drafts for verification. Combined effects, expression, specificity and PCR conditions require separate validation. A teaching CDS does not replace your real sequence.

Interpreting experimental analysis

Means, SD and WT differences are calculated from entered numbers. Their reliability depends on measurements, conditions, replicates and table completeness. Activity and stability are assessed separately. Teaching data are labelled as demonstrations.

In the pH section, the maximum is the observed point among tested values. It is not a proven global optimum or a prediction of mutation-induced shifts. Uploading a series does not automatically train a model.

Scope and rights

The service supports protein research planning. Results are not medical advice, diagnosis or authorisation for clinical use. Do not upload identifiable patient data.

These limitations explain what calculations mean without removing mandatory user rights, liability for technical defects or the obligation to supply paid deliverables.