Jee, S. | bioRxiv Science

bioRxiv2025

A DNA Sequence Imaging Approach to Predict Splice Sites Using Deep Learning

We present a novel approach to splice site prediction using image-based deep learning, comparing the established Frequency Chaos Game Representation (FCGR) with our proposed Dinucleotide Fixed Color Pattern (DFCP) technique. Applied to donor and acceptor splice site sequences from Arabidopsis thaliana and Homo sapiens, DFCP consistently outperforms FCGR in accuracy, precision, recall, and F1-score when using a ResNet50 model. Visualization techniques such as saliency maps and Grad-CAM further demonstrate that DFCP produces more localized and biologically interpretable activation patterns. These findings highlight the critical role of sequence visualization strategies in enhancing deep learning performance and interpretability in genomic analysis.

bioinformatics↗

bioRxiv2025

Frequency-Blended Diffusion Models for Synthetic Generation of Biologically Realistic Splice Site Sequences

We present a frequency-blended diffusion framework for generating biologically realistic splice site sequences. Our approach combines a U-Net-based denoising diffusion probabilistic model with conditional nucleotide frequency priors derived from real donor (5) and acceptor (3) splice site sequences in Arabidopsis thaliana and Homo sapiens. By guiding the generative process with position-specific empirical base frequencies, the model captures both local sequence motifs and long-range dependencies that are critical for realistic splice site representation. We evaluate the synthetic sequences through direct assessments (sequence logos, GC content, nucleotide conservation) and indirect functional tests using state-of-the-art splice site classifiers (SpliceRover, SpliceFinder, DeepSplicer, Spliceator). Our results show that frequency blending substantially improves motif conservation, compositional fidelity, and model transferability. This work establishes frequency-blended diffusion as a promising strategy for generating high-quality nucleotide sequences for modeling, benchmarking, and data augmentation in genomics research.

bioinformatics↗