iDHS-DMCAC

iDHS-DMCAC identifies DNase I hypersensitive sites (DHSs) in genomic sequences to enable analysis of regulatory DNA elements associated with gene expression and regulation.

Key Features:

• Statistical feature extraction: Uses the detrended moving-average cross-correlation (DMCA) coefficient descriptor to extract statistical features from DNA sequences.
• Dinucleotide property matrix: Constructs a 15-dimensional DNA dinucleotide property matrix to represent sequence properties.
• Feature vector construction: Derives a 105-dimensional feature vector for specified genomic windows from the dinucleotide property matrix.
• Class imbalance handling: Applies over-sampling techniques to address class imbalance in datasets.
• Classification algorithm: Employs support vector machine (SVM) algorithms for DHS classification.
• Performance evaluation: Uses rigorous cross-validation on benchmark datasets and reports improved accuracy and stability compared with existing models.

Scientific Applications:

• DHS identification: Identification of DNase I hypersensitive sites in genomic data.
• Regulatory element analysis: Analysis of regulatory DNA elements associated with gene expression and regulation.
• Genomic research: Support for high-throughput genomic studies investigating gene regulation and potential therapeutic targets.

Methodology:

Generate a 15-dimensional DNA dinucleotide property matrix, construct 105-dimensional feature vectors for genomic windows, compute DMCA coefficient descriptors, apply over-sampling techniques, train and test support vector machine classifiers, and evaluate performance by cross-validation on benchmark datasets.