Related Experiment Video
Updated: Jan 7, 2026
12:31
A Method for Selecting Structure-switching Aptamers Applied to a Colorimetric Gold Nanoparticle Assay
Published on: February 28, 2015
15.6K
Structure-enhanced deep learning accelerates aptamer selection for small molecule families like steroids.
Zibin Zhao1, Haosi Lin1, Hoi Ying Lau1
1Department of Chemical and Biological Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong SAR 999077, China.
Briefings in Bioinformatics
|December 22, 2025
Summary
Deep Learning-assisted SELEX (DL-SELEX) accelerates aptamer discovery by using variational autoencoders (VAEs) to design initial libraries and analyze results. This novel method significantly enhances aptamer affinity and reduces selection steps for challenging small molecules.
Area of Science:
- Biotechnology
- Computational Biology
- Molecular Biology
Background:
- Systematic Evolution of Ligands by EXponential enrichment (SELEX) is crucial for discovering high-affinity aptamers.
- Designing initial aptamer libraries and navigating vast sequence spaces are key challenges in SELEX.
- Current SELEX methods face limitations in efficiency and rational design for small-molecule aptamers.
Purpose of the Study:
- To introduce Deep Learning-assisted SELEX (DL-SELEX), a novel framework to enhance small-molecule aptamer discovery.
- To integrate deep learning for rational design of initial aptamer libraries and refine selection processes.
- To demonstrate the generalizability of DL-SELEX for structurally similar targets.
Main Methods:
- Developed a two-step framework, DL-SELEX, utilizing variational autoencoders (VAEs).
- Employed AptaVAE for tailored initial aptamer pool generation using transfer learning.
- Utilized AptaClux to identify high-performance aptamers from next-generation sequencing data by capturing consensus structural features.
Main Results:
- DL-SELEX yielded aptamers with up to 450-fold higher affinity for hydrocortisone and testosterone.
- Reduced the number of SELEX iterations by up to 80%.
- Demonstrated that deep learning models can be trained on structural commonalities for aptamer design.
Conclusions:
- DL-SELEX offers an effective and generalizable strategy to streamline aptamer discovery.
- This approach enables the de novo design of high-affinity aptamers for challenging small molecules.
- The integration of deep learning significantly advances SELEX workflows for small-molecule aptamer selection.
