Uncertainty Quantification and Flagging of Unreliable Predictions in Predicting Mass Spectrometry-Related Properties of Small Molecules Using Machine Learning DOI Open Access
Dmitriy D. Matyushin,

Ivan A. Burov,

Anastasia Yu. Sholokhova

и другие.

International Journal of Molecular Sciences, Год журнала: 2024, Номер 25(23), С. 13077 - 13077

Опубликована: Дек. 5, 2024

Mass spectral identification (in particular, in metabolomics) can be refined by comparing the observed and predicted properties of molecules, such as chromatographic retention. Significant advancements have been made predicting these values using machine learning deep learning. Usually, model predictions do not contain any indication possible error (uncertainty) or only one criterion is used for this purpose. The spread several models included ensemble, molecular similarity considered molecule most "similar" from training set, are that allow us to estimate uncertainty. Euclidean distance between vectors, calculated based on real-valued descriptors, assessment similarity. Another factor indicating uncertainty molecule's belonging clusters (data set clustering). Together, all three factors features model. Classification predict whether a prediction belongs worst 15% were obtained. area under receiver operating curve value range 0.73-0.82 tasks: retention indices gas chromatography, times liquid collision cross-sections ion mobility spectroscopy.

Язык: Английский

ULaMDyn: Enhancing Excited-State Dynamics Analysis Through Streamlined Unsupervised Learning DOI Creative Commons
Max Pinheiro, Matheus de Oliveira Bispo, Rafael S. Mattos

и другие.

Digital Discovery, Год журнала: 2025, Номер unknown

Опубликована: Янв. 1, 2025

The analysis of nonadiabatic molecular dynamics (NAMD) data presents significant challenges due to its high dimensionality and complexity. To address these issues, we introduce ULaMDyn, a Python-based, open-source package designed automate the unsupervised large datasets generated by NAMD simulations. ULaMDyn integrates seamlessly with Newton-X platform employs advanced reduction clustering techniques uncover hidden patterns in trajectories, enabling more intuitive understanding excited-state processes. Using photochemical fulvene as test case, demonstrate how efficiently identifies critical geometries transitions. offers streamlined, scalable solution for interpreting datasets. It is poised facilitate advances study across wide range systems.

Язык: Английский

Процитировано

1
Large-Scale Non-Adiabatic Dynamics Simulation Based on Machine Learning Hamiltonian and Force Field: The Case of Charge Transport in Monolayer MoS2 DOI

Bichuan Cao,

Jiawei Dong, Zedong Wang

и другие.

The Journal of Physical Chemistry Letters, Год журнала: 2025, Номер unknown, С. 4907 - 4920

Опубликована: Май 9, 2025

We present an efficient and reliable large-scale non-adiabatic dynamics simulation method based on machine learning Hamiltonian force field. The quasi-diabatic network (DHNet) is trained in the Wannier basis well-designed translation rotation invariant structural descriptors, which can effectively capture both local nonlocal environmental information. Using representative two-dimensional transition metal dichalcogenide MoS2 as illustration, we show that density functional theory (DFT) calculations of only ten structures are sufficient to generate training set for DHNet due high efficiency analysis orbital classification sampling interorbital couplings. demonstrates good transferability, thus enabling direct construction electronic matrices large systems. Compared with DFT calculations, significantly reduces computational cost by about 5 orders magnitude. By combining DeePMD field, successfully simulate electron transport monolayer up 3675 atoms 13475 levels using a state-of-the-art surface hopping method. mobility calculated be 110 cm2/(V s), agreement extensive experimental results range 3-200 s) during 2013-2023. Due performance, proposed methods have great potential applied study charge carrier wide material

Язык: Английский

Процитировано

0
Complementing Adiabatic and Nonadiabatic Methods To Understand Internal Conversion Dynamics in Porphyrin Derivatives DOI
Pavel S. Rukin, Mariagrazia Fortino, Deborah Prezzi

и другие.

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер unknown

Опубликована: Дек. 11, 2024

We analyze the internal conversion dynamics within

Язык: Английский

Процитировано

1
Uncertainty Quantification and Flagging of Unreliable Predictions in Predicting Mass Spectrometry-Related Properties of Small Molecules Using Machine Learning DOI Open Access
Dmitriy D. Matyushin,

Ivan A. Burov,

Anastasia Yu. Sholokhova

и другие.

International Journal of Molecular Sciences, Год журнала: 2024, Номер 25(23), С. 13077 - 13077

Опубликована: Дек. 5, 2024

Mass spectral identification (in particular, in metabolomics) can be refined by comparing the observed and predicted properties of molecules, such as chromatographic retention. Significant advancements have been made predicting these values using machine learning deep learning. Usually, model predictions do not contain any indication possible error (uncertainty) or only one criterion is used for this purpose. The spread several models included ensemble, molecular similarity considered molecule most "similar" from training set, are that allow us to estimate uncertainty. Euclidean distance between vectors, calculated based on real-valued descriptors, assessment similarity. Another factor indicating uncertainty molecule's belonging clusters (data set clustering). Together, all three factors features model. Classification predict whether a prediction belongs worst 15% were obtained. area under receiver operating curve value range 0.73-0.82 tasks: retention indices gas chromatography, times liquid collision cross-sections ion mobility spectroscopy.

Язык: Английский

Процитировано

0