PLANNER

PLANNER predicts eukaryotic DNA replication origins (ORIs) at species-specific and cell-specific resolution using multi-scale k-tuple sequence features and a DNABERT-based deep learning framework to identify and interpret sequence determinants.

Key Features:

• Species-Specific and Cell-Specific Predictions: Provides tailored ORI predictions at both species and cell-type levels for eukaryotic genomes.
• Multi-Scale k-tuple Sequences Input: Uses multi-scale k-tuple sequence representations to capture genomic patterns associated with ORI localization.
• Advanced Deep Learning Model: Employs the DNABERT pre-training model with transfer learning and ensemble learning strategies to train predictive models.
• Comparative Performance: The DNABERT-based approach yields higher predictive accuracy than iOri-Euk, Stack-ORI, and ORI-Deep according to reported evaluations.
• Interpretable Analysis Mechanism: Integrates an interpretable analysis component to reveal learned sequential determinants and relate them to biological function.

Scientific Applications:

• Cellular replication processes: Enables mapping of ORIs to study replication initiation sites and replication dynamics in eukaryotic cells.
• Gene expression regulation: Supports investigation of how ORI location and sequence context may influence local chromatin state and transcriptional regulation.
• Mutation-related diseases: Assists analysis of ORI-associated sequence features that may contribute to genome instability and disease-associated mutational patterns.

Methodology:

Multi-scale k-tuple sequences are processed by a DNABERT pre-trained deep learning framework that applies transfer learning and ensemble learning, and an interpretable analysis mechanism maps discovered sequential determinants to biological functions.