Effects of pulsed electric field parameters on anionic peptide migration during Electromembrane separation of a whey protein hydrolysate: A machine learning comprehensive study.

Abstract

Protein-rich byproducts such as whey represent valuable sources of bioactive peptides, but their selective separation remains a major challenge in sustainable bioprocessing. However, electrodialysis with ultrafiltration membrane (EDUF) offers a promising route for peptide fractionation, and recent evidence suggests that pulsed electric field (PEF) can modulate migration efficiency and selectivity. However, the mechanistic links between PEF parameters, peptide physicochemical characteristics and migration outcomes remain poorly resolved. This study evaluated six pulse/pause combinations (1 s/1 s, 5 s/1 s, 5 s/5 s, 10s/1 s, 10s/5 s and 10s/10s) during EDUF of a whey protein hydrolysate and combined peptidomics with machine learning-based regression tree analysis to identify peptide physicochemical characteristics governing their migration. The regression tree models (R = 49.0-74.9%) demonstrated hydrophobicity, leucine content, molecular weight and Kappa 1 as the most discriminant descriptors, with the lower prediction accuracy reflecting the complex interplay between peptide physicochemical properties and the dynamic interfacial conditions generated by PEF. After comprehensive analysis of the regression trees, three main mechanistic groups were defined and representative peptides were then tracked across all PEF conditions to reveal their migration behaviors. Group 1 peptides (no migration) did not migrate primarily by steric hindrance and electrostatic repulsion due to their high MW and/or high content in acidic residues preventing meaningful approach to the negatively charged PES 50 kDa UF membrane interface. Group 2 peptides (intermediate migration) migrating only when the time-averaged electric driving and the transiently favorable interfacial conditions created by PEF were jointly sufficient to overcome their moderate diffusive and hydrodynamic constraints combined with a diffusion boundary layer (DBL) state that remains partially destabilized. Group 3 peptides (high migration), small and highly hydrophobic, migrated efficiently across all conditions due to their low size and weak solvation, and in some cases through peptide-peptide hydrophobic interactions. Altogether, these results highlight that tuning PEF parameters, according to peptides physicochemical characteristics, enables fine control of peptide migration and selectivity in EDUF guiding the targeted recovery of bioactive fractions from complex hydrolysates.

AI evidence extraction

Main findings

In EDUF of a whey protein hydrolysate, machine learning regression trees identified hydrophobicity, leucine content, molecular weight, and Kappa 1 as key descriptors associated with peptide migration under pulsed electric field conditions. Peptides were grouped into no migration, intermediate migration, and high migration patterns, with migration behavior varying according to peptide properties and pulse/pause settings.

Outcomes measured

• Anionic peptide migration during electromembrane separation
• Migration efficiency and selectivity
• Peptide physicochemical descriptors associated with migration behavior

Limitations

• Prediction accuracy of regression tree models was limited (R = 49.0-74.9%)
• Complex interplay between peptide physicochemical properties and dynamic interfacial conditions reduced predictive performance
• Study focused on whey protein hydrolysate in an EDUF system, which may limit generalizability

{
    "study_type": "engineering",
    "exposure": {
        "band": null,
        "source": "other",
        "frequency_mhz": null,
        "sar_wkg": null,
        "duration": "Six pulse/pause combinations: 1 s/1 s, 5 s/1 s, 5 s/5 s, 10 s/1 s, 10 s/5 s, and 10 s/10 s during electrodialysis with ultrafiltration membrane (EDUF)."
    },
    "population": "Whey protein hydrolysate peptides",
    "sample_size": null,
    "outcomes": [
        "Anionic peptide migration during electromembrane separation",
        "Migration efficiency and selectivity",
        "Peptide physicochemical descriptors associated with migration behavior"
    ],
    "main_findings": "In EDUF of a whey protein hydrolysate, machine learning regression trees identified hydrophobicity, leucine content, molecular weight, and Kappa 1 as key descriptors associated with peptide migration under pulsed electric field conditions. Peptides were grouped into no migration, intermediate migration, and high migration patterns, with migration behavior varying according to peptide properties and pulse/pause settings.",
    "effect_direction": "mixed",
    "limitations": [
        "Prediction accuracy of regression tree models was limited (R = 49.0-74.9%)",
        "Complex interplay between peptide physicochemical properties and dynamic interfacial conditions reduced predictive performance",
        "Study focused on whey protein hydrolysate in an EDUF system, which may limit generalizability"
    ],
    "evidence_strength": "low",
    "confidence": 0.9499999999999999555910790149937383830547332763671875,
    "peer_reviewed_likely": "yes",
    "keywords": [
        "pulsed electric field",
        "electrodialysis with ultrafiltration membrane",
        "EDUF",
        "whey protein hydrolysate",
        "peptide migration",
        "machine learning",
        "regression tree",
        "peptidomics",
        "hydrophobicity",
        "molecular weight"
    ],
    "suggested_hubs": []
}

Comments

Log in to comment.

No comments yet.