ORI-Deep

ORI-Deep predicts origins of replication (ORIs) from genomic DNA sequences to support studies of DNA replication mechanisms.

Key Features:

• Deep Neural Network Utilization: ORI-Deep employs a long short-term memory (LSTM) network to model sequential genomic DNA patterns indicative of replication origins.
• Feature Vector Construction: ORI-Deep constructs input feature vectors using statistical moments derived from sequence data.
• Comprehensive Validation Techniques: Performance was assessed using self-consistency checks, 10-fold cross-validation, jackknife tests, and independent set evaluation.
• High Accuracy Scores: Reported accuracies are 0.977 (self-consistency), 0.948 (10-fold cross-validation), 0.976 (jackknife), and 0.977 (independent set).

Scientific Applications:

• ORI prediction in genomic research: Enables identification of ORIs to study DNA replication mechanisms in genomic datasets.
• Comparative and evolutionary analyses: Applied across various eukaryotic species to provide insights into evolutionary biology.
• Gene expression regulation studies: Aids exploration of relationships between replication origin location and gene expression regulation.
• Cellular growth and structural development research: Supports investigation of cellular growth patterns and structural development at the molecular level.

Methodology:

Constructs feature vectors using statistical moments from sequence data, employs a long short-term memory (LSTM) network for modeling, and evaluates performance via self-consistency checks, 10-fold cross-validation, jackknife tests, and independent set testing.