Table 1.
Comparison of the Properties of Previous Deep Learning Models Predicting Gene Expression with GC-MERGE
| Computational study | Inputs |
Interpretation |
|||
|---|---|---|---|---|---|
| Local histone marks | Additional features (e.g., promoter sequence) | Long-range interactions | Feature-level interpretation | Edge-level interpretation | |
| DeepChrome | X | X | |||
| AttentiveChrome | X | X | |||
| Xpresso | X | X | |||
| DeepExpression | X | X | X | ||
| GC-MERGE | X | X | X | X | |
The proposed method integrates local and long-range regulatory interactions, capturing the underlying 3D genomic spatial structure as well as highlighting both the critical node-level (histone modifications) and edge-level (genomic interactions) features.
GC-MERGE, Graph Convolutional Model for Epigenetic Regulation of Gene Expression.