Ajay Yadav, PhD

Development of Highly Specific Aptamers Against Carbendazim Using Machine Learning-Guided Computational Strategies

Ajay Yadav1 and Hariprasad P.1* 1 Environmental Biotechnology Lab Centre for Rural Development and Technology Indian Institute of Technology Delhi Hauz Khas, New Delhi 110016, India

Aptamers are short, single-stranded oligonucleotides that can selectively recognize diverse targets, including small molecules, proteins, toxins, hormones, and even entire microbial cells, with high affinity and specificity. However, the conventional SELEX process for aptamer discovery is labor-intensive, time-consuming, and costly. To address these challenges, we developed a machine learning-based framework to accelerate aptamer screening and predict aptamer-small molecule interactions with high precision. Our approach employs a two-layer stacked model: the first layer (RF_1) uses global features of aptamers and small molecules to estimate binding probabilities and generate a meta-score, while the second layer (RF_2) integrates this meta-score with local aptamer features to refine predictions. This hierarchical design enhances prediction accuracy and robustness. The stacked model achieved 86.7% accuracy and an ROC-AUC of 0.94, reliably distinguishing binders from non-binders. Applying the model to carbendazim, we screened aptamers from HT-SELEX libraries (0th and 8th cycles) as well as mutated carbendazim-specific aptamers (CZ1, CZ2, CZ5, CZ6, ZC12, CZ13), yielding binding probability ranges of 0.12-0.96, 0.30-0.96, and 0.19-0.94, respectively. The top 10 candidates from each dataset were further validated via in vitro dissociation constant (Kd) measurements, where selected aptamers exhibited nanomolar binding affinities. These results demonstrate that our model effectively predicts aptamer-carbendazim interactions, enabling rapid and cost-efficient identification of high-affinity binders. Importantly, the framework is broadly applicable to other small molecules, offering a scalable strategy for accelerating aptamer discovery.