A systematic review of deep learning chemical language models in recent era DOI Creative Commons

Hector Flores-Hernandez,

Emmanuel Martínez-Ledesma

Journal of Cheminformatics, Journal Year: 2024, Volume and Issue: 16(1)

Published: Nov. 18, 2024

Discovering new chemical compounds with specific properties can provide advantages for fields that rely on materials their development, although this task comes at a high cost in terms of complexity and resources. Since the beginning data age, deep learning techniques have revolutionized process designing molecules by analyzing from representations molecular data, greatly reducing resources time involved. Various approaches been developed to date, using variety architectures strategies, order explore extensive discontinuous space, providing benefits generating properties. In study, we present systematic review offers statistical description comparison strategies utilized generate through techniques, utilizing metrics proposed Molecular Sets (MOSES) or Guacamol. The study included 48 articles retrieved query-based search Scopus Web Science 25 citation search, yielding total 72 articles, which 62 correspond language models molecule generation other 10 graph representations. Transformers, recurrent neural networks (RNNs), generative adversarial (GANs), Structured Space State Sequence (S4) models, variational autoencoders (VAEs) are considered main used set articles. addition, transfer learning, reinforcement conditional most employed biased model exploration space regions. Finally, analysis focuses central themes representation, databases, training dataset size, validity-novelty trade-off, performance unbiased models. These were selected conduct graphical representation tests. resulting reveals challenges, advantages, opportunities field over past four years.

Language: Английский

Assessment of Students Use of Generative Artificial Intelligence: Prompting Strategies and Prompt Engineering in Chemistry Education DOI Creative Commons
Sebastian Tassoti

Journal of Chemical Education, Journal Year: 2024, Volume and Issue: 101(6), P. 2475 - 2482

Published: May 22, 2024

The rapid integration of generative artificial intelligence (AI) into educational settings prompts an urgent examination its efficacy and the strategies that students employ to harness potential. This study focuses on preservice chemistry teachers use AI for chemistry-specific problem-solving task completion. We found there is a prevalent reliance copy-pasting tactics in initial prompting approaches, need guidance improve their abilities. By implementing "Five S" framework, we explore evolution student resultant satisfaction with AI-generated responses. Our findings indicate that, while initially struggle nuances effective prompting, adoption structured frameworks significantly enhances perceived quality answers. research sheds light current state among but also underscores importance targeted refine interaction academic contexts. In particular, suggest critical engagement methodological prompt engineering maximize benefits technologies.

Language: Английский

Citations

24
Image and data mining in reticular chemistry powered by GPT-4V DOI Creative Commons
Zhiling Zheng,

Zhiguo He,

Omar Khattab

et al.

Digital Discovery, Journal Year: 2024, Volume and Issue: 3(3), P. 491 - 501

Published: Jan. 1, 2024

The integration of artificial intelligence into scientific research opens new avenues with the advent GPT-4V, a large language model equipped vision capabilities.

Language: Английский

Citations

18
Trust, Explainability and AI DOI Creative Commons
Sam Baron

Philosophy & Technology, Journal Year: 2025, Volume and Issue: 38(1)

Published: Jan. 8, 2025

Abstract There has been a surge of interest in explainable artificial intelligence (XAI). It is commonly claimed that explainability necessary for trust AI, and this why we need it. In paper, I argue some notions it plausible indeed condition. But these kinds are not appropriate AI. For thus conclude AI matters.

Language: Английский

Citations

3
InterDIA: Interpretable Prediction of Drug-induced Autoimmunity through Ensemble Machine Learning Approaches DOI

Lina Huang,

P. L. Liu,

Xiaojie Huang

et al.

Toxicology, Journal Year: 2025, Volume and Issue: unknown, P. 154064 - 154064

Published: Jan. 1, 2025

Language: Английский

Citations

0
Exploring high-performance viscosity index improver polymers via high-throughput molecular dynamics and explainable AI DOI Creative Commons

Rui Zhou,

Luyao Bao, Weifeng Bu

et al.

npj Computational Materials, Journal Year: 2025, Volume and Issue: 11(1)

Published: March 1, 2025

Language: Английский

Citations

0
ACtriplet: An improved deep learning model for activity cliffs prediction by integrating triplet loss and pre-training DOI Creative Commons

Xinxin Yu,

Yimeng Wang, Long Chen

et al.

Journal of Pharmaceutical Analysis, Journal Year: 2025, Volume and Issue: unknown, P. 101317 - 101317

Published: April 1, 2025

Language: Английский

Citations

0
COLOR: A Compositional Linear Operation-Based Representation of Protein Sequences for Identification of Monomer Contributions to Properties DOI
Akash Pandey, Wei Chen, Sinan Keten

et al.

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: April 24, 2025

The properties of biological materials like proteins and nucleic acids are largely determined by their primary sequence. Certain segments in the sequence strongly influence specific functions, but identifying these segments, or so-called motifs, is challenging due to complexity sequential data. While deep learning (DL) models can accurately capture sequence-property relationships, degree nonlinearity limits assessment monomer contributions a property─a critical step key motifs. Recent advances explainable AI (XAI) offer attention gradient-based methods for estimating monomeric contributions. However, primarily applied classification tasks, such as binding site identification, where they achieve limited accuracy (40-45%) rely on qualitative evaluations. To address limitations, we introduce DL model with interpretable steps, enabling direct tracing Inspired masking technique commonly used vision natural language processing domains, propose new metric (I) quantitative analysis datasets mainly containing distinct anticancer peptides (ACP), antimicrobial (AMP), collagen. Our exhibits 22% higher explainability than gradient attention-based state-of-the-art models, recognizes motifs (RRR, RRI, RSS) that significantly destabilize ACPs, identifies AMPs 50% more effective converting non-AMPs AMPs. These findings highlight potential our guiding mutation strategies designing protein-based biomaterials.

Language: Английский

Citations

0
Cross-disciplinary perspectives on the potential for artificial intelligence across chemistry DOI Creative Commons
Austin M. Mroz, Annabel R. Basford, Friedrich Hastedt

et al.

Chemical Society Reviews, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

We offer ten diverse perspectives exploring the transformative potential of artificial intelligence (AI) in chemistry, highlighting many challenges we face, and offering strategies to address them.

Language: Английский

Citations

0
An Ensemble Cascade Forest‐Based Framework for Multi‐Omics Drug Response and Synergy Prediction DOI Creative Commons
Ruijiang Li,

Binsheng Sui,

Dongjin Leng

et al.

Advanced Intelligent Systems, Journal Year: 2024, Volume and Issue: unknown

Published: June 9, 2024

The obscure drug response continues to be a limiting factor for accurate cures cancer. Next generation sequencing technologies have propelled the pharmacogenomic studies with characterized large panels of cancer cell line at multi‐omics level. However, sufficient integration data and efficient prediction synergy still remain challenge. To address these problems, ECFD is designed, an ensemble cascade forest‐based framework that predicts using five types omics data. Experimental results show significant advantages model over existing models. best feature extraction determined superiorities robust stability in face new small samples are highlighted. In addition, methodological highlights explainability model, mechanisms resistance combination treatment strategies based on explainable analyses biological networks. sum, may facilitate evaluation speculation potential therapies personalized precision treatment.

Language: Английский

Citations

1
Systematic bibliometric and visualized analysis of research hotspots and trends on the application of artificial intelligence in glaucoma from 2013 to 2022 DOI
Chun Liu,

Luyao Wang,

Keyu Zhu

et al.

International Journal of Ophthalmology, Journal Year: 2024, Volume and Issue: 17(9), P. 1731 - 1742

Published: Aug. 20, 2024

To conduct a bibliometric analysis of research on artificial intelligence (AI) in the field glaucoma to gain comprehensive understanding current state and identify potential new directions for future studies.

Language: Английский

Citations

1