A groundbreaking advancement in the field of nucleic acid aptamer discovery has recently emerged, spearheaded by researchers Weihong Tan, Xiaohong Fang, and Tao Bing from the Hangzhou Institute of Medical Sciences at the Chinese Academy of Sciences. Their innovative approach leverages machine learning techniques to decode the complex secondary structures of nucleic acid aptamers from single-round screening data. This cutting-edge methodology remarkably bypasses the traditionally lengthy iterative enrichment processes, directly extracting detailed structural information necessary for optimizing high-affinity aptamers. Published as an open-access research article in CCS Chemistry, this study proposes a transformative paradigm in aptamer research, enhancing both speed and precision in aptamer identification and optimization.
Nucleic acid aptamers are short, single-stranded oligonucleotides that fold into intricate three-dimensional conformations, allowing them to bind with high specificity and affinity to various target molecules. Despite the power of SELEX (Systematic Evolution of Ligands by EXponential enrichment) in generating candidate aptamers, elucidating their functional secondary structures that mediate target recognition has remained a formidable challenge. Conventional structural determination methods, including electron microscopy, nuclear magnetic resonance (NMR), and X-ray crystallography, are not only resource-intensive but often fail to resolve the dynamic and heterogenous structures characteristic of aptamer-target complexes. Consequently, optimizing and truncating aptamer sequences to enhance binding efficiencies has been constrained by limited structural insights.
Addressing these longstanding hurdles, the research team has pioneered a sophisticated machine learning framework that integrates unsupervised autoencoder clustering with deep learning algorithms to dissect the core sequence elements within large pools of aptamer candidates obtained from a single screening round. This approach marks a departure from traditional iterative enrichment, enabling the identification of conserved sequence motifs and corresponding secondary structural features that are crucial for target binding. Through this strategy, the molecular architecture underlying aptamer function can be inferred computationally, providing a blueprint for rational design and refinement without extensive experimental trial-and-error.
The authors first applied their methodology to the screening dataset of aptamers targeting the CD8 protein, a critical cell surface marker. Utilizing deep learning to analyze sequence families within the single-round library, they uncovered a prevalent core sequence “GTGAGGAGCTTGAAA” despite the highly heterogeneous sequence background. Significantly, conventional multiple sequence alignment methods were inadequate in extracting these short motifs amidst the low homology environment, underscoring the superiority of the machine learning approach in resolving subtle sequence patterns pertinent to function.
To empirically validate the computational core sequence findings, the team synthesized a candidate library embedding the critical partial motif (5′-AGCTTGAAA-3′) and subjected it to RE-SILEX, a selection method. Astonishingly, all newly identified aptamers—over 20,000 in total—contained the predicted core sequences, attesting to the robustness of single-round screening combined with machine learning analysis. This provided a powerful proof of concept that the approach not only identifies biologically relevant motifs but also guides subsequent aptamer enrichment and design.
Taking the analysis further, the researchers developed a machine learning-based algorithm to interrogate the secondary structures formed by the core sequences in the fixed region. Statistical evaluation revealed that approximately 62.4% of these sequences formed stem-loop structures critical for molecular recognition, while the remainder adopted diverse conformations. Among the stem-loop forming aptamers, the sequence “GTGA” predominated within multi-branched loops and stem regions, suggesting a consensus binding motif. Detailed quantification and base distribution profiling confirmed a shared secondary structure among aptamers, revealing how specific folding patterns confer target specificity and affinity.
Informed by these insights, rational truncation and optimization strategies were applied to aptamers derived from the RE-SILEX pool, yielding sequences with significantly enhanced binding affinities. This optimization process capitalized on machine learning-driven structural knowledge to discard redundant nucleotides and concentrate on functionally critical motifs, achieving a remarkable increase in aptamer performance. Such advancements demonstrate the potential to streamline the aptamer development pipeline from initial discovery to functional application.
The universality of this single-round machine learning technique was demonstrated through analysis of a separate fibrogenic target, fibroblast activation protein (FAP). Here, the method identified a highly conserved core sequence hypothesized to form a G-quadruplex secondary structure, stabilized by hairpin formations at terminal regions. This finding was pivotal, as it illustrated the adaptability of the approach to differing structural motifs beyond stem-loops, extending its scope to more complex, non-canonical nucleic acid topologies. Successive truncation and affinity optimization on FAP aptamers further validated the broad applicability of the framework.
In subsequent validation experiments focusing on the CD8 protein, the research team observed that over three-quarters of sequences bore the identified core motif and shared coherent secondary structural elements. The truncated, optimized aptamers exhibited affinity enhancements exceeding tenfold while preserving target specificity, even within complex cell environments. Remarkably, the structural characterization also facilitated the design of split-type aptamers and opened avenues for de novo sequence generation, heralding a new era in synthetic biology toolkits employing deep learning for functional nucleic acid design.
Beyond immediate applications, this study highlights a transformative shift in aptamer research paradigms—from iterative enrichment and experimental structural elucidation to data-driven, computationally empowered discovery. By revealing that single-round libraries inherently contain rich structural and functional information, the work fundamentally challenges the notion that extensive screening cycles are necessary. The deployment of high-throughput sequencing coupled with advanced machine learning thus enables rapid, precise decoding of the nucleic acid folding landscapes that dictate target binding.
The implications of this research extend beyond aptamers into broader biomolecular and therapeutic contexts. The machine learning algorithms developed can be adapted for exploring non-coding RNA interactions and modeling RNA-protein complexes with unprecedented resolution. Moreover, by facilitating AI-driven virtual screening platforms for nucleic acid ligands, this approach paves the way for accelerated drug discovery pipelines and precision diagnostics, ultimately catalyzing the development of next-generation nucleic acid-based therapeutics tailored to personalized medicine.
Supported by prominent funding agencies such as the National Natural Science Foundation of China, the Zhejiang Provincial “Pioneer” and “Leading Goose” R&D Program, and the Strategic Priority Research Program of the Chinese Academy of Sciences, this research exemplifies the convergence of computational innovation and molecular biology. Its publication in CCS Chemistry underscores the increasing role of interdisciplinary methodologies in driving forward chemical sciences at the international forefront.
As the scientific community continues to harness the power of artificial intelligence in molecular design, the synergy of machine learning with high-throughput sequencing heralds a new frontier in aptamer technology. This integration not only refines our understanding of structure-function relationships but also empowers rational engineering of nucleic acids with bespoke functionalities. Such advances will accelerate the translation of aptamers from bench to bedside, enhancing their utility in diagnostics, therapeutics, and beyond.
Subject of Research: Not applicable
Article Title: Single-Round Aptamer Discovery Empowered by Machine Learning: Revealing Structure–Function Principles of Target Binding
News Publication Date: 7-Jan-2026
Web References:
https://www.chinesechemsoc.org/journal/ccschem
http://dx.doi.org/10.31635/ccschem.025.202506736
Image Credits: CCS Chemistry
Keywords
Machine learning
