LedPred

LedPred predicts cis-regulatory modules (CRMs) by training support vector machine (SVM) classifiers that integrate position-specific scoring matrices, ChIP-seq data, and conservation scores to score and classify DNA sequences.

Key Features:

• Integrated data utilization: Incorporates position-specific scoring matrices, ChIP-seq data, and conservation scores as heterogeneous features for CRM prediction.
• SVM-based supervised classification: Uses support vector machines trained on annotated CRM data to distinguish regulatory from non-regulatory sequences.
• Feature-to-sequence mapping workflow: Maps diverse feature types to sequence instances to enable model training and prediction based on known CRM annotations.
• Sequence scoring: Produces predictive scores for unknown DNA sequences to identify candidate regulatory elements.
• Benchmark performance: Demonstrated superior performance in predicting regulatory sequences on Drosophila and mouse datasets compared with similar SVM-based tools.

Scientific Applications:

• CRM prediction in model organisms: Identification of cis-regulatory modules in Drosophila and mouse genomic datasets.
• Regulatory element discovery: Prioritization of candidate regulatory elements that control gene expression for studies of genetic networks, development, and disease.

Methodology:

Supervised classification using support vector machines trained on annotated CRM data, integrating position-specific scoring matrices, ChIP-seq data, and conservation scores, with mapping of heterogeneous features to sequences and scoring of unknown DNA sequences.