3D chemical structures allow robust deep learning models for retention time prediction

Zaretckii, Mark; Bashkirova, Inga; Osipenko, Sergey; Kostyukevich, Yury; Nikolaev, Eugene; Popov, Petr

doi:10.1039/d2dd00021k

Cited by 5 publications

(4 citation statements)

References 43 publications

(90 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The prepared database was used for construction of 11 ML models after the generation of fingerprint descriptors using RDKit AllChem software. The RDKit AllChem software was chosen due to it being a freely available tool with the ability to generate a broad range of molecular descriptors, including structural, topological, electronic, and thermodynamic properties of molecules. , The molecular descriptors used in this study included molecular weight, number of rotatable bonds, topological polar surface area (TPSA), and Morgan fingerprints with radii of 2 and 1024 bits. K -fold cross-validation was used as the primary validation method with k set to 5.…”

Section: Resultsmentioning

confidence: 99%

Machine Learning-Based Approach to Developing Potent EGFR Inhibitors for Breast Cancer─Design, Synthesis, and In Vitro Evaluation

Nada,

Gul,

Elkamhawy

et al. 2023

ACS Omega

View full text Add to dashboard Cite

The epidermal growth factor receptor (EGFR) is vital for regulating cellular functions, including cell division, migration, survival, apoptosis, angiogenesis, and cancer. EGFR overexpression is an ideal target for anticancer drug development as it is absent from normal tissues, marking it as tumor-specific. Unfortunately, the development of medication resistance limits the therapeutic efficacy of the currently approved EGFR inhibitors, indicating the need for further development. Herein, a machine learning-based application that predicts the bioactivity of novel EGFR inhibitors is presented. Clustering of the EGFR smallmolecule inhibitor (∼9000 compounds) library showed that Nsubstituted quinazolin-4-amine-based compounds made up the largest cluster of EGFR inhibitors (∼2500 compounds). Taking advantage of this finding, rational drug design was used to design a novel series of 4-anilinoquinazoline-based EGFR inhibitors, which were first tested by the developed artificial intelligence application, and only the compounds which were predicted to be active were then chosen to be synthesized. This led to the synthesis of 18 novel compounds, which were subsequently evaluated for cytotoxicity and EGFR inhibitory activity. Among the tested compounds, compound 9 demonstrated the most potent antiproliferative activity, with 2.50 and 1.96 μM activity over MCF-7 and MDA-MB-231 cancer cell lines, respectively. Moreover, compound 9 displayed an EGFR inhibitory activity of 2.53 nM and promising apoptotic results, marking it a potential candidate for breast cancer therapy.

show abstract

Section: Resultsmentioning

confidence: 99%

Machine Learning-Based Approach to Developing Potent EGFR Inhibitors for Breast Cancer─Design, Synthesis, and In Vitro Evaluation

Nada,

Gul,

Elkamhawy

et al. 2023

ACS Omega

View full text Add to dashboard Cite

show abstract

“…Recently, small molecule retention time (SMRT) dataset ( Domingo-Almenara et al 2019 ) containing 80 038 molecules was released to the public, stimulating deep learning-based retention time prediction methods, such as DLM ( Domingo-Almenara et al 2019 ), DNNpwa ( Ju et al 2021 ), and 1D-CNN ( Fedorova et al 2022 ). More deep learning-based methods, such as GNN-RT ( Yang et al 2021 ), CPORT ( Zaretckii et al 2022 ), MPNN ( Osipenko et al 2022 ), and Blender ( García et al 2022 ), apply transfer learning ( Weiss et al 2016 ) to predict the retention times in specific chromatographic separation systems. These methods alleviate the limitation of small training data by pre-training the neural networks on SMRT and further reusing some parameters in the pre-trained networks.…”

Section: Introductionmentioning

confidence: 99%

RT-Transformer: retention time prediction for metabolite annotation to assist in metabolite identification

Xue,

Wang,

et al. 2024

Bioinformatics

View full text Add to dashboard Cite

Motivation Liquid chromatography retention times prediction can assist in metabolite identification, which is a critical task and challenge in non-targeted metabolomics. However, different chromatographic conditions may result in different retention times for the same metabolite. Current retention time prediction methods lack sufficient scalability to transfer from one specific chromatographic method to another. Results Therefore, we present RT-Transformer, a novel deep neural network model coupled with graph attention network and 1D-Transformer, which can predict retention times under any chromatographic methods. First, we obtain a pre-trained model by training RT-Transformer on the large small molecule retention time dataset containing 80038 molecules, and then transfer the resulting model to different chromatographic methods based on transfer learning. When tested on the small molecule retention time dataset, as other authors did, the average absolute error reached 27.30 after removing not retained molecules. Still, it reached 33.41 when no samples were removed. The pre-trained RT-Transformer was further transferred to 5 datasets corresponding to different chromatographic conditions and fine-tuned. According to the experimental results, RT-Transformer achieves competitive performance compared to state-of-the-art methods. In addition, RT-Transformer was applied to 41 external molecular retention time datasets. Extensive evaluations indicate that RT-Transformer has excellent scalability in predicting retention times for liquid chromatography and improves the accuracy of metabolite identification. Availability and Implementation The source code for the model is available at https://github.com/01dadada/RT-Transformer.The web server is available at https://huggingface.co/spaces/Xue-Jun/RT-Transformer Supplementary information Supplementary data are available at Bioinformatics online.

show abstract

“…The representation of molecules in 3D grids of voxels encoding their atoms has been used to train deep learning neural networks. 18,19,20,21 Electrostatic potentials generated by point charges localized at the atom centers -another way of representing the 3D structure of molecules -has long since been used to establish relationships with molecular properties, namely in 3D-QSAR, 22 or to assess molecular similarity. 23 Direct 3D representations can be used for data sets of pre-aligned structures (e.g.…”

Section: Introductionmentioning

confidence: 99%

GUIDEMOL: a Python graphical user interface for molecular descriptors based on RDKit

Aires-de-Sousa

2023

Preprint

View full text Add to dashboard Cite

GUIDEMOL is a Python computer program based on the RDKit software to process molecular structures and calculate molecular descriptors with a graphical user interface using the tkinter package. It can calculate descriptors already implemented in RDKit as well as grid representations of 3D molecular structures using the electrostatic potential or voxels. The GUIDEMOL app provides an easy access to RDKit tools for chemoinformatics users with no programming skills and can be adapted to calculate other descriptors or to trigger other procedures. A CLI is also provided for the calculation of grid representations. The source code is available at https://github.com/jairesdesousa/guidemol

show abstract

3D chemical structures allow robust deep learning models for retention time prediction

Abstract: Chromatographic retention time (RT) is a powerful characteristic used to identify, separate, or rank molecules in a mixture. With accumulated RT data, it becomes possible to develop deep learning approaches...

Cited by 5 publications

References 43 publications

Machine Learning-Based Approach to Developing Potent EGFR Inhibitors for Breast Cancer─Design, Synthesis, and In Vitro Evaluation

Machine Learning-Based Approach to Developing Potent EGFR Inhibitors for Breast Cancer─Design, Synthesis, and In Vitro Evaluation

RT-Transformer: retention time prediction for metabolite annotation to assist in metabolite identification

GUIDEMOL: a Python graphical user interface for molecular descriptors based on RDKit

Contact Info

Product

Resources

About