G: Gibbs free energy is a thermodynamic potential (G = H − TS) that, at constant temperature and pressure, represents the maximum non-expansion work a system can perform.

Δ_rG: The Gibbs free energy change of a reaction under a given set of (e.g., physiological) metabolite concentrations, which determines thermodynamic feasibility.

Δ_rG°: The Gibbs free energy change of a reaction under standard conditions, reflecting the relative thermodynamic stability of reactants and products.

The actual and standard reaction free energies are related by ΔrG = ΔrG° + RT ln Q, where R is the universal gas constant, T is the absolute temperature, and Q is the reaction quotient determined by reactant and product concentrations.

dGbyG web is a web-based platform that uses graph neural networks to predict the standard Gibbs free energy change (Δ_rG°) of biochemical reactions. The name stands for “Δ_rG° predicted by Graph neural networks.”

The platform provides two main services:

• Search GEMs: Browse 100+ genome-scale metabolic models (GEMs) with precomputed Δ_rG° values.
• Predict Δ_rG°: Compute Δ_rG° for custom reactions using our AI model.

dGbyG is a graph neural network–based tool for predicting the standard reaction Gibbs free energy change (Δ_rG°) of metabolic reactions.

1. Molecular graphs: Each metabolite is represented as an atom–bond graph with rich chemical features.
2. GNN prediction: The model learns thermodynamic representations from these graphs and predicts Δ_fG° for metabolites. Δ_rG° is then computed as the stoichiometry-weighted sum of Δ_fG° values of substrates and products.
3. Methodological advantage: Unlike group-contribution methods that rely on predefined chemical groups, dGbyG models atoms and bonds directly, improving robustness and generalization when training data are limited or reaction mechanisms are unseen.
4. Condition parameters: Users can specify pH, ionic strength (I), pMg, and electrical potential (V) to reflect physiological environments.
5. Output: The predicted Δ_rG° together with an uncertainty estimate.

Compared with the best existing approaches, dGbyG reduces the median prediction error on the validation set from 5.33 kJ/mol to 4.11 kJ/mol, while increasing reaction coverage in the human genome-scale metabolic model Recon3D from 64.14% to 71.22%.

For more details, please refer to: Unraveling principles of thermodynamics for genome-scale metabolic networks using graph neural networks.

SD reports the estimated model uncertainty (kJ/mol) from a bootstrap-based ensemble. It captures how much the predicted Δ_rG° varies when the model is trained on different noise-perturbed versions of the training data.

• How we compute it: We assume experimental Δ_rG° measurements have Gaussian uncertainty. In each bootstrap run, we perturb training labels according to their estimated measurement variability, train an independent model, and repeat this process 100 times. For a query reaction, we compute Δ_rG° for each model and report the standard deviation across the 100 predictions. A variability index α is also used to weight training points in the loss so that noisier measurements have less influence.
• How to interpret it: A smaller SD indicates higher agreement across the ensemble (higher confidence).

Our GEM library currently includes 115 genome-scale metabolic models:

• BiGG Models (107 GEMs, excluding Recon3D): sourced from BiGG Models.
• Metabolic Atlas (7 GEMs): sourced from the Metabolic Atlas GEM repository. (For these 7 models, we retain only reactions and metabolites annotated with BiGG IDs.)
• Recon3D (1 GEM): the human Recon3D model from the latest release by Elizabeth Brunk et al. (SBRG/Recon3D).

You can search for:

• Models: By model name (e.g., "Mouse-GEM", "iML1515", "Recon3D")
• Reactions: By reaction BiGG ID or name (e.g., "HEX1", "Hexokinase")
• Metabolites: By metabolite BiGG ID (with compartment, e.g., "g6p_c") or BiGG Universal ID (without compartment, e.g., "g6p")
• Genes: By gene ID or name (e.g., "3948_AT2", "LDHC")

For each model compartment (e.g., cytosol, mitochondrion, lysosome), we set the parameters used in Δ_rG° calculations—pH, ionic strength (I), pMg, and electrical potential (E)—as follows:

1. Literature: Values reported for the relevant species and compartment.
2. BioNumbers: If not available in the literature, we use typical compartment values from BioNumbers.
3. Closest-species: If neither source provides a value, we adopt conservative values from closely related species.

The selected parameters are shown on the model page under “Compartment Conditions.”

We support multiple compound identifier formats:

• SMILES: Simplified Molecular Input Line Entry System (e.g., CCO for ethanol)
• InChI: International Chemical Identifier (e.g., InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3 for ethanol)
• InChI Key: Hashed version of InChI (e.g., LFQSCWFLJHTTHZ-UHFFFAOYSA-N for ethanol)
• KEGG ID: KEGG compound identifiers (e.g., C00031 for glucose)
• MetaNetX ID: MetaNetX database identifiers (e.g., MNXM7381 for glucose)
• ChEBI ID: Chemical Entities of Biological Interest identifiers (e.g., 17234 for glucose)
• PubChem CID: PubChem compound IDs (e.g., 702 for ethanol)
• HMDB ID: Human Metabolome Database identifiers (e.g., HMDB0000122 for glucose)
• Compound Name: Common chemical names (e.g., phosphate, acetate)
• BiGG Universal ID: BiGG identifiers without compartment suffixes (e.g., succ, not succ_c)
• Mixed Identifiers: A single query can combine formats using TYPE:ID (e.g., CHEBI:37736).

Recommendation: Use SMILES or InChI for the best accuracy and reliability.

Rules:

• Use = (with spaces) to separate reactants and products.
• Use + between compounds, and place coefficients in front (e.g., 2 H2O).

Examples:

You can either choose a preset for common compartments or define your own conditions.

Presets: Standard, Cytosol, Mitochondria, Lysosome, Golgi, etc. (each preset fills pH, I, pMg, E, and T)

Custom conditions:

• pH (0–14)
• Ionic strength (0–1 M)
• pMg (0–14)
• Electrical potential (−1 to 1 V)

Yes. Batch mode can process up to 100 reactions in a single run.

• Paste one equation per line and, if needed, set global pH / I / pMg / E.
• Download the results as a CSV file.

Δ_rG° values are reported at 25 °C (298.15 K).

• Most experimental thermodynamic data, as well as our model training, are standardized at 298 K.
• Reliable temperature adjustment would require reaction-specific ΔH and ΔS (ΔG = ΔH − TΔS), and often heat-capacity terms, which are not available at the genome scale.