Advanced in silico analysis of yeast-derived bioactive peptides

In food science and biotechnology sectors, the accurate evaluation of the safety, bioactivity, and pharmacological potential of novel bioactive peptides is essential for the development of functional foods, nutraceuticals, and therapeutic agents. This thesis evaluated the prediction of several biological properties of fifteen bioactive peptides, obtained from yeast through enzymatic hydrolysis of yeast biomass, using computer-based techniques. The research goal focused on finding potential bioactivities for therapeutical and nutraceutical uses of food waste-derived peptides, while demonstrated the effectiveness of multi-tool computational screening for novel peptide research. The research used five different computational tools (i.e., ToxinPred, CPPpred, SwissADME, admetSAR and PeptideRanker) to evaluate peptide potential toxicity levels, cell penetration abilities, pharmacokinetic properties, and bioactivity. The ToxinPred analysis predicted that 13 out of 15 peptides could display non-toxic results, which indicates yeast-derived sequences carry a natural safety feature. The analysis results showed that P1, P3 and P6 peptides have strong cell-penetrating abilities but other peptides demonstrated moderate or weak penetration abilities. The SwissADME analysis results showed that most peptides have high gastrointestinal absorption rates, appropriate lipophilicity, and acceptable TPSA values which could make them suitable for oral delivery. The admetSAR tool confirmed the safety of these peptides because it showed all peptides, except for P3 and P11, as non-carcinogenic and non-mutagenic. PeptideRanker identified several highly bioactive peptides, particularly P1, P3, P6, and P14, suggesting strong functional potential. By integrating results across all tools, three peptides, i.e., P1, P6, and P14, emerged as the most promising candidates due to their combined safety, high predicted bioactivity, favorable ADMET characteristics, and low allergenic potential. This research shows that complete computer-based predictive systems may help scientists to find new potentially bioactive peptides, speeding up the experimental evaluation processes toward the most promising peptide candidates.