Phycocyanin (PC) is the principal natural blue pigment used in functional beverages, but it rapidly loses color and aggregates under acidic conditions (pH ~3). Experimental screening of stabilizers is costly and combinatorially intractable. Here we develop a chemoinformatics framework that learns from three rounds of commissioned screening (49 compounds, 8% hit rate) to predict stabilizer efficacy directly from molecular structure. A LightGBM classifier built from 10 RDKit descriptors and 11 domain-expert charge/polymer features attained a leave-one-out AUC of 0.88, outperforming a single-feature charge-density baseline (AUC 0.75). SHAP analysis identified polyphosphate identity, effective negative-charge density and chain length as the dominant features, together accounting for more than 80% of the model's feature attribution. A complementary chemistry-prior heuristic encoding anion-type priors reached an AUC of 0.96, indicating that explicit chemical knowledge captures information that small datasets cannot readily recover from descriptors alone. Virtual screening of 30 generally recognized as safe (GRAS) food additives nominated the pyrophosphate family - led by sodium hexametaphosphate (SHMP) and tetrasodium pyrophosphate (TSPP) - and the heuristic additionally suggested sodium phytate (IP6), which the descriptor model under-ranked. Experimental validation at pH 3 and 46 degrees C for 7 days confirmed SHMP 2:1 (78.1 +/- 11.3% color retention), TSPP 2:1 (54.1 +/- 10.6%) and IP6 1:1 (52.7 +/- 9.0%), while a ternary IP6 + STPP combination reached 83.8 +/- 11.9%, surpassing all single-component formulations. Zeta-potential measurements indicated a predominantly electrostatic origin for the protection (Pearson r = -0.82 between zeta-potential and CR620; n = 24; p = 1 x 10-6). The framework, dataset and code are released to accelerate stabilizer discovery for other acid-sensitive food colorants.
law, l., Chuang, K., Luo, L.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 7
- Comments 0