Selection of aptamers is crucial iterative process which is significant for many applications from biosensors to biomedical applications. The process of selecting aptamers is called systematic evolution of ligands by exponential enrichment (SELEX). This process is costly, time consuming and labour-intensive. In this communication we present T_SELEX API together with its algorithms and application. This program is an autonomous robust python package specifically engineered for designing aptamers and screening of RNA aptamers against macromolecular targets. The tool is evaluated from generation of sequence library of RNA aptamers, multiscale folding or secondary structure predictions, multiscale tertiary structure predictions, virtual screening and detailed analysis. The application was tested by screening a generated aptamer library against HIV-1 protease. The source code of the T_SELEX tool is available on GitHub [https://github.com/CMCDD/T_SELEX] and can be used directly following instructions on the readme file.

While databases are emerging across various domains, from small molecules to genomics and proteins, aptamer databases remain scarce, if not entirely absent. Such databases could serve as a comprehensive resource for advancing research, innovation, and the applications of aptamer technology across multiple fields. This advancement would likely lead to improvements in healthcare, environmental monitoring, and biotechnology. Furthermore, the establishment of aptamer databases would facilitate molecular modelling and machine learning, opening doors to further advancements in understanding and utilizing aptamers. Against this backdrop, in this study, we present and benchmark the Base Randomization Algorithm (BRA) as a potential solution to the scarcity of aptamer databases. Through statistical analysis, we examine key factors such as minimum free energy (MFE), base compositions, and base arrangements. Notably, sequences generated using the BRA exhibit a Gaussian distribution pattern. We also examine the details of how each base within a sequence is chosen using mathematical principles, ensuring that the sequences are valid and optimized statistically. Additionally, we explore how the length of the randomized generated sequences can affect the folding of their structures at both the secondary and tertiary levels. Based on composition analysis, we propose that the base mean of the dataset can be approximated as ${\overline{x}}_B\approx P\left(x\right) \times N$, for dataset of sequences with the same length and ${\overline{x}}_B\approx P\left(x\right) \times M,$ where $M$ is the median and $N$ the mean, for a dataset with randomized length that follows a Gaussian distribution.