Call for Papers

We invite submissions for our upcoming Special Track on Advancing Molecular Analysis with AI at The 21st International Conference Advanced Data Mining and Applications 2025 (ADMA 2025).

Background

Artificial intelligence (AI) is transforming molecular analysis, revolutionizing how we identify, characterize, and design chemical compounds. Traditional methods for molecular structure elucidation rely on labor-intensive spectral interpretation and heuristic-based computational techniques, often requiring significant expertise and time to extract meaningful insights from complex data. Such approaches can be prone to human error and are limited by the scope of pre-existing knowledge and computational power. As the complexity of molecular systems increases, there is an urgent need for more efficient, scalable, and accurate techniques to meet the growing demands of modern research.

AI-driven approaches, such as graph neural networks (GNNs), variational autoencoders (VAEs), generative adversarial networks (GANs), and diffusion models, provide a powerful data-driven paradigm for molecular analysis. These techniques facilitate automated spectral annotation, de novo molecular generation, and predictive modeling of chemical properties, transforming the way molecular research is conducted. Such advancements are particularly critical in fields like drug discovery, metabolomics, proteomics, and materials science, where the need for rapid and accurate molecular characterization is essential for driving innovation and accelerating breakthroughs.

Despite these promising developments, several challenges remain in applying AI to　molecular analysis, which include

Data scarcity and heterogeneity plague AI models, as high-quality, annotated spectral datasets (e.g., for rare metabolites or novel polymers) remain limited.

Sparse, noisy, and instrument-dependent datasets often lack standardized annotations.

The stochastic nature of molecular fragmentation in MS/MS data introduces uncertainty in spectral interpretation.

Generalization from known to unknown compounds without sacrificing accuracy is still difficult due to the vast combinatorial chemical space.

Generating chemically valid, synthesizable, and biologically relevant molecules requires additional constraints and physics-informed modeling.

The "black-box" nature of many AI systems raises concerns about interpretability, particularly in regulated fields like pharmaceuticals, where validation is paramount.

Addressing these challenges with recent advancements in AI, such as large language models, will enhance AI's ability to extract meaningful representations from spectral and structural data. AI's potential benefits in molecular analysis are profound: faster and more accurate compound identification, automated drug design, efficient retrosynthesis planning, and the discovery of novel materials with tailored properties.

This special session aims to bring together researchers from computational chemistry, cheminformatics, AI, and analytical sciences to discuss recent advances in AI-driven molecular analysis, focusing on applications in mass spectrometry (MS), nuclear magnetic resonance (NMR), molecular docking, and drug discovery. We welcome innovative AI methodologies that address challenges such as noisy spectral data, fragmentation pattern prediction, small molecule generation, and multi-modal molecular analysis.

Scope

We invite submissions related (but not limited) to the following topics:

• AI for Spectral Data Interpretation
  • Deep learning for MS/MS, NMR, and IR spectroscopy analysis
  • AI-driven spectral deconvolution and peak annotation
  • Generative models for spectrum-to-structure prediction
• De Novo Molecular Generation & Property Prediction
  • Generative AI models for molecular design
  • AI-guided chemical space exploration and inverse molecular design
  • Optimization of chemical properties, such as drug-likeness and bioactivity
• Molecular Generation with Multiview or Multimodal
  • Molecular generation from SMILES, graphs, SELFIES, and/or images
  • Molecular generation from omics data
  • Molecular generation from text
• Graph-Based Approaches for Molecular Representation
  • Graph neural networks (GNNs) for chemical structure learning
  • Self-supervised learning for molecular embeddings
  • Multi-modal learning for integrating spectral, structural, and bioactivity data
• Machine Learning for Chemical Reaction and Synthesis Prediction
  • AI for predicting reaction mechanisms and retrosynthesis
  • AI-assisted experimental design and synthetic pathway optimization
  • Large language models (LLMs) for automated molecular discovery
• Benchmarking, Explainability, and Generalization of AI Models in Chemistry
  • Creation of AI benchmarks for molecular analysis
  • Model interpretability and explainable AI (XAI) in cheminformatics
  • Transfer learning and domain adaptation for diverse chemical datasets

This session targets researchers and practitioners from AI, cheminformatics, computational chemistry, bioinformatics, and analytical sciences, including:

• Academics in AI for chemistry, cheminformatics, machine learning and data science
• Industry professionals in pharmaceuticals, materials science, and analytical chemistry
• Researchers in proteomics, metabolomics, foodomics, forensics, environmental science and structural biology
• Experts in medical diagnostics and health technology

Formatting Guidelines

We welcome English-language papers containing original and unpublished contributions to the fields of data mining and related areas. Manuscripts should adhere to the LNAI (Lecture Notes in Artificial Intelligence) format. For the template and detailed instructions on LNCS style, please refer to Springer's Author Instructions. Papers should adhere to the main conference guidelines, ensuring they do not exceed 15 pages in LNAI format. Submissions undergo a double-blind review process for ADMA2025. This means:

• Author identities and affiliations remain undisclosed to reviewers throughout the review process.
• Authors must prepare and submit blinded manuscripts that conceal author and affiliation information. Specific guidelines are outlined below.
• Both authors and reviewers are expected to make sincere efforts to prevent accidental de-blinding of any submission.
• All submitted papers must adhere to the following rules. Non-compliance may result in automatic rejection.

Submission Guidelines

Authors are invited to submit original research papers, case studies, and technical reports aligned with the theme of Data Science: Foundations and Applications. Submissions should adhere to the conference's formatting guidelines and be submitted through the CMT Submission System. All submissions will undergo a rigorous peer-review process to ensure quality and relevance. When submitting your manuscript, please choose the "Special Session Track" option and select the area of "Special Session on Advancing Molecular Analysis with AI."

Important Dates

• Paper Submission Deadline: 20th May, 2025 (AoE)
• Notification of Acceptance: 27th July, 2025 (AoE)
• Camera-Ready Paper Due: 10 August, 2025 (AoE)
• Conference Dates: 22nd-24th Ocober, 2025 (AEST)

Session Chairs

• Lan Du, Monash University, Australian
• Geoff Webb, Monash University, Australian
• Manabu Sugimoto, Kumamoto University, Japan

Program Committee

• Wray Buntine, VinUniversity, Vietnam
• Zhiming Li, Fudan University, China
• Changyou Chen, State University of New York, USA
• Ying Ma, Harbin Institute of Technology, China
• Jianpin Gou, Southwest University, China
• Xuyun Zhang, Macquarie University, Australia