In Silico Prediction of Potential Bile Salt Export Pump Inhibitors

Sebastian Schieferdecker presents a dataset supporting the development of a consensus machine learning classification model for predicting inhibitors of the bile salt export pump (BSEP). The model was validated by in vitro measurements, correctly flagging eight out of 11 predicted compounds. The dataset is hosted on figshare and was last updated on 2026-04-24.

Use Cases

• Training machine learning models for BSEP inhibition prediction based on pharmacophore features.
• Prioritizing compounds for in vitro BSEP inhibition testing based on model predictions.
• Assessing drug safety risk during development by comparing predicted IC50 values against the 25 μM advisory threshold.
• Defining the applicability domain of a predictive model for BSEP inhibitors.

Strengths

• Model validation shows eight out of 11 predicted compounds were correctly flagged.
• Dataset is focused on a specific, high-impact safety target (BSEP) relevant to drug-induced liver injury.
• The work provides a concrete IC50 advisory threshold of 25 μM for risk assessment.

Limitations

• Dataset size is very small at 37.3 KB, suggesting limited scope.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Sebastian Schieferdecker

Collection Method

In silico modeling using pharmacophore models and consensus machine learning, validated with in vitro measurements.

Freshness

Last updated 2026-04-24 09:44:45; freshness should be verified.