This thesis explores the integration of machine learning (ML) with molecular dynamics (MD), analyzing how ML can overcome the limitations of traditional simulations by improving their efficiency and accuracy. Innovative methods such as GDML for force field generation, ML-driven coarse-grained models, and protein motion analysis using AlphaFold are discussed. The dual role of MD as a data generator for training ML models is also highlighted, with practical examples such as the StrEAMM project. This combined ML-MD approach opens new perspectives in computational chemistry and (as explored in this thesis) drug development.
Questa tesi esplora il rapporto tra il machine learning (ML) e la dinamica molecolare (MD), analizzando come l’apprendimento automatico possa superare le limitazioni delle simulazioni tradizionali, migliorandone efficienza e accuratezza. Vengono descritti metodi innovativi come il GDML per generare campi di forze, l'applicazione del ML in modelli coarse-grained e nell’analisi dei moti proteici con AlphaFold. La tesi evidenzia inoltre il ruolo duale della MD come fornitore di dati per l’addestramento di modelli ML, illustrando esempi concreti come il progetto StrEAMM. L’approccio integrato ML-MD apre nuove prospettive nella chimica computazionale e (come esplorato in questa tesi) anche nello sviluppo di farmaci.
Machine learning e dinamica molecolare
GIOVANAZZI, MATTEO
2024/2025
Abstract
This thesis explores the integration of machine learning (ML) with molecular dynamics (MD), analyzing how ML can overcome the limitations of traditional simulations by improving their efficiency and accuracy. Innovative methods such as GDML for force field generation, ML-driven coarse-grained models, and protein motion analysis using AlphaFold are discussed. The dual role of MD as a data generator for training ML models is also highlighted, with practical examples such as the StrEAMM project. This combined ML-MD approach opens new perspectives in computational chemistry and (as explored in this thesis) drug development.| File | Dimensione | Formato | |
|---|---|---|---|
|
Giovanazzi_Matteo.pdf
accesso aperto
Dimensione
1.3 MB
Formato
Adobe PDF
|
1.3 MB | Adobe PDF | Visualizza/Apri |
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/20.500.12608/89637