CEPZ

CEPZ predicts DNase I hypersensitive sites (DHSs) by integrating sequence composition and physicochemical properties to identify cis-regulatory elements in chromatin.

Key Features:

• Predictive performance: Reports a Matthews correlation coefficient (MCC) of 0.7740 and an overall prediction accuracy of 91.13% for DHS identification.
• Feature integration: Combines composition information and physicochemical properties of genomic sequences as input features.
• Feature selection: Applies a boosting algorithm to select features most indicative of DHSs.
• Dinucleotide analysis: Examines dinucleotide properties and identifies dinucleotides with significant distributional differences between DHS and non-DHS samples.

Scientific Applications:

• Regulatory region identification: Predicts regions of open chromatin (DHSs) that serve as indicators of cis-regulatory elements for decoding gene regulatory networks.
• Epigenetics and gene expression studies: Locates DHSs to support analyses of epigenetic modifications and gene expression patterns.
• Genetic disease and comparative genomics: Assists investigation of mechanisms underlying genetic diseases and analysis of regulatory landscapes across human and other complex genomes.

Methodology:

Integrates sequence composition and physicochemical properties, employs a boosting algorithm for feature selection, and analyzes dinucleotide distributional differences to build the DHS predictive model.