ConSReg

ConSReg predicts condition-specific transcriptional regulators and their target genes by integrating protein–DNA interaction and open chromatin data with machine learning models to analyze single-cell and condition-specific gene expression such as abiotic stress.

Key Features:

• Integration with Genomic Data: Integrates protein–DNA interaction and open chromatin region data across multiple eukaryotic species for condition-specific regulatory inference.
• Machine Learning Approach: Employs a novel machine learning methodology to predict regulatory genes from datasets including single-cell gene expression and abiotic stress treatments.
• Performance Metrics: Achieved an average area under the receiver operating characteristic curve (auROC) of 0.84 in Arabidopsis, representing approximately 23.5–25% higher performance than enrichment-based approaches.

Scientific Applications:

• Transcription Factor Identification: Identifies transcription factors that regulate differentially expressed genes under specific conditions such as abiotic stress and single-cell gene expression scenarios.
• Validation Across Datasets: Validated on independent plant nitrogen response datasets, producing better rankings of correct transcription factors in 61.7% of cases—three times more effectively than existing plant-specific methodologies.
• Single-Cell RNA-seq Analysis: Applied to Arabidopsis single-cell RNA sequencing data to identify candidate regulatory genes implicated in processes such as cell wall formation.

Methodology:

Implemented as a Python package that integrates protein–DNA interaction and open chromatin data into predictive machine learning models, analyzes single-cell gene expression and abiotic stress treatment datasets, evaluates performance using auROC and comparisons to enrichment-based methods, and validates results on independent plant nitrogen response datasets.