PlantDeepSEA

PlantDeepSEA predicts regulatory effects of genomic variants in plants using deep learning models trained on chromatin-profiling data to assess chromatin accessibility across tissues in multiple plant species, including four key crops.

Key Features:

• Variant Effector: Predicts how sequence variants influence chromatin accessibility across tissues in six plant species.
• Sequence Profiler: Performs in silico saturated mutagenesis to identify high-impact sites within sequences, including cis-regulatory elements.
• Model Training: Uses deep learning models trained on chromatin-profiling data.
• Variant Prioritization: Ranks and prioritizes genomic variants based on predicted regulatory impact to aid identification of causal variants.
• Performance: Models validated with area under receiver operating characteristic curve (AUC) values ranging from 0.93 to 0.99.

Scientific Applications:

• Mechanistic insight: Provides insights into mechanisms by which genomic variants exert regulatory effects.
• Tissue-specific regulation: Predicts variant impacts across multiple tissues to inform studies of gene regulation and expression patterns.
• Experimental prioritization: Prioritizes variants for experimental validation in functional genomics studies.

Methodology:

Employs deep learning models trained on chromatin-profiling data, applies in silico saturated mutagenesis for sequence-level effect estimation, and reports validation performance as AUCs ranging from 0.93 to 0.99.