DL-ML-Hybrid

DL-ML-Hybrid integrates deep neural networks (DNNs) and non-deep learning classifiers to predict transgenerational differential DNA methylated regions (DMRs) and epimutations for epigenetic and transgenerational inheritance studies.

Key Features:

• Hybrid Approach: Employs a two-stage pipeline where a deep neural network (DNN) extracts molecular features from biological sequence data.
• Feature Extraction and Classification: Uses DNN-derived features as input to a non-deep learning classifier for classification of DMRs and epimutations.
• Computational Efficiency: Uses a less complex DNN for feature selection to reduce computational expense relative to full-scale deep learning models.
• Application in Epigenetics: Predicts environmentally responsive transgenerational differential DNA methylated regions (DMRs) and sperm epimutations associated with environmental toxicants.
• Cross-Species Prediction Capability: Applied to rat and human genomes to enable cross-species prediction of potential DMRs.

Scientific Applications:

• Epigenetic Research: Identification of transgenerational epimutations induced by environmental toxicants to study epigenetic inheritance and gene–environment interactions.
• Genomic Studies: Genome-wide prediction of DMRs across rat and human genomes for comparative genomics and analyses of genetic regulation patterns.

Methodology:

A DNN is trained for feature extraction using environmental toxicant-induced epigenetic transgenerational inheritance sperm epimutations from rat genome DNA sequences, and a non-deep learning classifier is trained on the DNN-derived features and applied to predict DMRs across genomes including human genomic data.