MOGT Model Performance on a BioGRID-Excluded Negative Sample Set

Model performance metrics for the MOGT model evaluated on an alternative negative sample set that excludes interactions from the BioGRID database. The dataset was authored by Jiafang Li and last updated on May 29, 2026. Performance was assessed using AUROC, F1-score, and AUPRC on a hold-out test set, with model selection conducted via 5-fold cross-validation.

Use Cases

• Benchmarking model performance based on AUROC, F1-score, and AUPRC metrics
• Comparing model robustness based on evaluation using a hold-out test set
• Assessing model selection strategies based on the use of 5-fold cross-validation
• Analyzing the impact of negative sample construction based on the exclusion of BioGRID interactions

Strengths

• Performance evaluation includes three standard metrics: AUROC, F1-score, and AUPRC
• Model selection was conducted using a 5-fold cross-validation methodology
• Dataset is openly licensed under CC-BY-4.0
• File is provided in the widely compatible XLSX format

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

figshare

Collection Method

Likely contains results from a computational experiment evaluating a machine learning model.

Freshness

Last updated 2026-05-29 18:00:11; freshness should be verified