March 3, 2026Open Access

Interpretable prediction of DNA replication origins in S. cerevisiae using attention-based motif discovery

Key Points

Biologically relevant motifs were identified, advancing the understanding of dna replication origins and their sequence composition.
The analysis combined attention maps with MEME, enhancing feature discovery while minimizing manual extraction efforts.
Using DNABERT enables robust prediction of yeast replication origins across datasets with various complexities.
This approach supports improved interpretability in machine learning models within genomics, potentially benefiting broader applications.

Abstract

In a living cell, DNA replication begins at multiple genomic sites, called replicationorigins. Identifying these origins and their underlying base sequence compositionis crucial for understanding replication process. Existing machine learningmethods for origin prediction often require labor-intensive feature engineering orlack interpretability. Here, we employ DNABERT to predict yeast replication originsand uncover sequence motifs by combining attention maps with MEME, aclassical bioinformatics tool. Our approach eliminates manual feature extractionand identifies biologically relevant motifs across datasets of varying complexity.This work advances interpretable machine learning in genomics, offering a potentiallygeneralizable framework for origin prediction and motif discovery.

Interpretable prediction of DNA replication origins in S. cerevisiae using attention-based motif discovery

Key Points

Abstract

Cite This Study