DLAM: Deep Learning Model Predictions for Essential Proteins

DLAM is a deep learning framework for predicting essential proteins, integrating domain composition, subcellular localization, orthology, gene expression, and a weighted protein-protein interaction network. The dataset contains model performance results on the DIP and BioGRID datasets, achieving strong discrimination and ranking metrics. It was authored by Shunxian Zhou and last updated on April 14, 2026.

Use Cases

• Benchmarking new essential protein prediction models based on reported ROC-AUC and AP scores.
• Training or validating classifiers using multi-source biological features like domain composition and subcellular localization.
• Analyzing the impact of class imbalance on model performance metrics such as F1-score, precision, and recall.
• Studying the integration of heterogeneous biological signals with network topology for protein essentiality scoring.

Strengths

• Model performance is quantitatively reported with metrics like ROC-AUC, AP, accuracy, precision, recall, and F-measure.
• Comparative evaluation includes four recent deep learning methods (TCBB2021, EPGAT, BMC2022, ACDMBI) on the larger BioGRID dataset.
• Evaluation used stratified five-fold cross-validation, suggesting a robust validation protocol.

Limitations

• Row count and specific column-level documentation are absent; field semantics must be inferred after download.
• The dataset is 116.3 KB, indicating a very limited scope, likely containing only summary performance metrics rather than raw training data.

Provenance

Source

figshare, author Shunxian Zhou

Collection Method

Results from a deep learning model (DLAM) integrating multi-source biological information.

Time Range

null

Freshness

Last updated 2026-04-14 22:01:27; freshness should be verified.

Geography

null