Bioactive peptides are short protein fragments. They stay inactive while still part of the parent protein. Once released, they can gain antioxidant, anti-inflammatory, antimicrobial or antihypertensive activity. Recovering them from dairy by-products, such as mastitic milk, turns discarded biomass into a source of functional ingredients. This brings the food chain into a circular-economy perspective. This thesis continues a previous MSc project on milk-derived bioactive peptides and the related study on peptide-nanoparticle complexes (Tonolo et al., 2025). That earlier work identified four bioactive peptides from bovine milk proteins: G-14-K, I-14-K, E-14-R and N-12-L, which were also evaluated for their antioxidant and anti-inflammatory activity in a cellular model. Building on that work, I assembled a seven-tool in silico pipeline. It screens these four peptides together with 22 further peptides from the same dataset, already annotated as active in the BIOPEP database. The pipeline covers toxicity, cell-penetrating potential, anti-inflammatory and antioxidant potential, allergenicity, small-molecule ADMET (absorption, distribution, metabolism, excretion and toxicity) properties and general bioactivity probability. Three of the originally planned tools were no longer reachable, so I replaced them with documented substitutes: CellPPD, AllerCatPro 2.0 and PepAnno. All 26 peptides came back non-toxic and were largely unable to cross a cell membrane on their own. Several returned anti-inflammatory and antioxidant scores consistent with the activity already measured experimentally in the related published study (Tonolo et al., 2025). Allergenicity screening flagged 17 of the 22 peptides for sequence similarity to known bovine milk allergens. admetSAR, on the other hand, proved unsuitable for molecules of this size. As a further, self-contained contribution, I designed and built a similarity-search tool, delivered as a standalone desktop application. It predicts peptide bioactivity by local sequence alignment against reference databases, rather than through a dedicated classifier. Applied to the same 26 peptides, it correctly recovered the known literature identity of most casein-derived candidates and reported no strong match for the four non-casein peptides, as expected. Extended to the remaining 295 peptides of the same LC-MS/MS dataset that still carry no BIOPEP-assigned activity, it returned a high-confidence prediction for 78 of them. Together, these results sketch a prioritization strategy for the experimental validation of bioactive peptides recovered from dairy-processing waste.
I peptidi bioattivi sono brevi frammenti proteici. Non presentano alcuna attività finché sono parte della proteina di origine. Una volta liberati, possono acquisire diversi effetti benefici, come per esempio proprietà antiossidante, antinfiammatoria, antimicrobica o antipertensiva. Recuperarli da scarti della filiera lattiero-casearia, come il latte di vacche affette da mastite, trasforma una biomassa altrimenti scartata in una fonte di ingredienti funzionali. Questa tesi è la prosecuzione di un precedente lavoro di tesi magistrale sui peptidi bioattivi e lo studio correlato sui nanocomplessi peptide-nanoparticella (Tonolo et al., 2025). In quest'ultimo lavoro sono stati identificati quattro peptidi bioattivi a partire da proteine del latte bovino: G-14-K, I-14-K, E-14-R e N-12-L, i quali sono stati analizzati in modelli cellulari per le loro possibili attività antiossidanti ed antinfiammatorie. Partendo da questo lavoro, ho costruito una pipeline in silico basata su sette software bioinformatici. In questo elaborato di tesi sono stati analizzati G-14-K, I-14-K, E-14-R e N-12-L contestualmente ad altri 22 peptidi dello stesso dataset, già identificati come attivi nel database BIOPEP. La pipeline valuta tossicità, potenziale di penetrazione cellulare, capacità antinfiammatoria ed antiossidante, allergenicità, proprietà ADMET (assorbimento, distribuzione, metabolismo, escrezione e tossicità) e qual è la probabilità generale di bioattività. Tre degli strumenti previsti inizialmente non erano più disponibili online; quindi, li ho sostituiti con alternative documentate: CellPPD, AllerCatPro 2.0 e PepAnno. Tutti i 26 peptidi sono risultati non tossici, e in gran parte incapaci di attraversare la membrana cellulare. Diversi hanno mostrato promettenti punteggi sulla capacità antinfiammatoria e antiossidante coerenti con l'attività già misurata sperimentalmente nello studio pubblicato correlato (Tonolo et al., 2025). Lo screening di allergenicità ha segnalato 17 dei 22 peptidi per similarità di sequenza con noti allergeni del latte bovino. admetSAR, invece, si è rivelato inadatto a molecole di queste dimensioni. Come ulteriore contributo, ho progettato e realizzato uno strumento di ricerca per similarità, consegnato come applicazione desktop indipendente. Predice la bioattività di un peptide tramite allineamento di sequenza contro database di riferimento, anziché tramite un classificatore dedicato. Applicato agli stessi 26 peptidi, ha correttamente ritrovato l'identità di letteratura nota della maggior parte dei candidati caseinici, e non ha riportato corrispondenze forti per i quattro peptidi non caseinici, come atteso. Esteso ai restanti 295 peptidi dello stesso dataset LC-MS/MS ancora privi di un'attività assegnata da BIOPEP, ha restituito una predizione ad alta confidenza per 78 di essi. Nel complesso, questi risultati delineano una strategia di prioritizzazione per la validazione sperimentale di peptidi bioattivi recuperati dagli scarti della lavorazione del latte.
Development of a Bioinformatics Pipeline for the Characterization of Bioactive Milk Peptides
MESCHIARI, LORENZO
2025/2026
Abstract
Bioactive peptides are short protein fragments. They stay inactive while still part of the parent protein. Once released, they can gain antioxidant, anti-inflammatory, antimicrobial or antihypertensive activity. Recovering them from dairy by-products, such as mastitic milk, turns discarded biomass into a source of functional ingredients. This brings the food chain into a circular-economy perspective. This thesis continues a previous MSc project on milk-derived bioactive peptides and the related study on peptide-nanoparticle complexes (Tonolo et al., 2025). That earlier work identified four bioactive peptides from bovine milk proteins: G-14-K, I-14-K, E-14-R and N-12-L, which were also evaluated for their antioxidant and anti-inflammatory activity in a cellular model. Building on that work, I assembled a seven-tool in silico pipeline. It screens these four peptides together with 22 further peptides from the same dataset, already annotated as active in the BIOPEP database. The pipeline covers toxicity, cell-penetrating potential, anti-inflammatory and antioxidant potential, allergenicity, small-molecule ADMET (absorption, distribution, metabolism, excretion and toxicity) properties and general bioactivity probability. Three of the originally planned tools were no longer reachable, so I replaced them with documented substitutes: CellPPD, AllerCatPro 2.0 and PepAnno. All 26 peptides came back non-toxic and were largely unable to cross a cell membrane on their own. Several returned anti-inflammatory and antioxidant scores consistent with the activity already measured experimentally in the related published study (Tonolo et al., 2025). Allergenicity screening flagged 17 of the 22 peptides for sequence similarity to known bovine milk allergens. admetSAR, on the other hand, proved unsuitable for molecules of this size. As a further, self-contained contribution, I designed and built a similarity-search tool, delivered as a standalone desktop application. It predicts peptide bioactivity by local sequence alignment against reference databases, rather than through a dedicated classifier. Applied to the same 26 peptides, it correctly recovered the known literature identity of most casein-derived candidates and reported no strong match for the four non-casein peptides, as expected. Extended to the remaining 295 peptides of the same LC-MS/MS dataset that still carry no BIOPEP-assigned activity, it returned a high-confidence prediction for 78 of them. Together, these results sketch a prioritization strategy for the experimental validation of bioactive peptides recovered from dairy-processing waste.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_Lorenzo_Meschiari_1237522.pdf
Accesso riservato
Dimensione
1.67 MB
Formato
Adobe PDF
|
1.67 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/117898