A Novel DNA Classification Experiment: Spatial Transcriptomics analysis for human Monkeypox DNA-Motifs with Kolmogorov–Arnold Networks

Authors : Selçuk Yazar

Pages : 839-851

Doi:10.24012/dumf.1537079

View : 60 | Download : 111

Publication Date : 2024-12-23

Article Type : Research Paper

Abstract :Spatial Transcriptomics(ST) has emerged as a powerful tool for understanding gene expression patterns across different regions of a tissue or organism. It is crucial for disease research and developing new therapies. It allows for the measurement of gene expression across specific, localized areas of a tissue slide, though it does so with limited throughput. Yet, the data produced by ST technologies are characteristically noisy, high-dimensional, sparse, and multi-modal, encompassing elements like histological images and count matrices. Existing methods for analyzing ST data, which often rely on traditional statistical or machine learning techniques, have proven inadequate in many cases due to challenges like scale, multi-modality, and the inherent limitations of spatially-resolved data, including spatial resolution, sensitivity, and gene coverage. To address these specific challenges, researchers have turned to deep learning-based models. In this study, we present a novel approach to transcriptomics analysis using Kolmogorov-Arnold Networks (KANs), a state-of-the-art deep learning model to predict regional origin of monkeypox transcriptomic sample. By leveraging the ability of KANs to learn and represent complex, non-linear functions, we aim to uncover intricate spatial patterns of gene expression and gain insights into the underlying biological processes. Study’s analysis focuses on two distinct regions, America and Asia, and employs a KAN-based classifier. The results demonstrate the promising performance of KANs in this context, with a precision of 0.45 and a recall of 0.93 for the America region, indicating a strong ability to correctly identify samples from this region. Findings indicate that predicting the regional transcriptome of monkeypox from DNA motifs could facilitate image-based screening for phylogenetic analyses.
Keywords : Mekânsal Transkriptomik, Kolmogorov-Arnold Ağları, Derin Öğrenme, Gen Örüntüleri, Bölgesel Sınıflandırma