ADMETlab 3.0 quickstart

You want a broad ADMET profile for a compound or library — covering absorption, distribution, metabolism, excretion, and a wide toxicity panel. This walks you from entering SMILES to reading the results in five steps: open the tool, add your compounds, choose an endpoint bundle, submit, and interpret the color-coded output.

Quick ADMET profiling with .bloom* platform

ADMETlab 3.0, published in Nucleic Acids Research (2024), provides comprehensive ADMET prediction across 119 endpoints using a multi-task Directed Message Passing Neural Network (DMPNN) architecture trained on over 400,000 data entries. This latest version introduces 31 new endpoints compared to v2.0, including nephrotoxicity, neurotoxicity, ototoxicity, and hematotoxicity predictions.

Step-by-step ADMET-Lab 3.0 workflow

Common ADMET profiling scenarios

Lead series evaluation

1. Upload analog series from medicinal chemistry
2. Run comprehensive ADMET profiling
3. Identify compounds with optimal PK properties
4. Generate optimization guidance based on property gaps

Virtual screening follow-up

1. Import hit compounds from virtual screening campaigns
2. Assess ADMET properties for drug development potential
3. Flag compounds with poor solubility or high toxicity risk
4. Prioritize compounds for experimental follow-up

Safety assessment

1. Select preclinical candidates from optimization campaigns
2. Run extended toxicity profiling
3. Evaluate cardiovascular and hepatotoxicity risks
4. Generate safety reports for project teams

Multi-tool validation

1. Run ADMET-Lab 3.0 predictions on compound library
2. Compare results with ADMET-AI predictions
3. Cross-validate important safety and PK assessments
4. Use consensus results for confident decision making

Quality control and validation

Input validation checklist

Compound preparation:

• Valid SMILES syntax (use external validators if needed)
• Neutral form preferred (remove counterions when appropriate)
• Reasonable molecular weight (typically <1000 Da)
• Drug-like chemical space (within training domain)

File format verification:

• CSV with SMILES in first column, optional IDs in second
• No special characters in compound identifiers
• Consistent SMILES representation across compounds
• UTF-8 encoding for international character support

Result quality assessment

Prediction reliability:

• Review confidence scores for critical predictions
• Check applicability domain warnings
• Validate against known compounds when available
• Consider experimental verification for key results

Batch processing validation:

• Ensure all compounds processed (check for failures)
• Verify consistent property ranges for similar compounds
• Review preprocessing results for standardization
• Confirm model version consistency across runs

Troubleshooting common issues

Compound processing errors

• Check SMILES validity using online validation tools
• Ensure compounds are in neutral form (remove salts if needed)
• Verify molecular weight is within acceptable ranges
• Remove unusual elements not supported by models

Prediction failures

• Review compound structures for unusual functional groups
• Check if compounds are too large (>1000 Da)
• Consider if molecules are outside chemical training domain
• Split large batches into smaller jobs for processing

Performance optimization

• Reduce batch size for complex molecules
• Choose appropriate endpoint bundle for your needs
• Use preprocessing to standardize compound representations
• Monitor job progress and retry failed predictions

Result interpretation challenges

• Understand risk category definitions for each endpoint
• Consider prediction confidence when making decisions
• Cross-reference with other ADMET tools when possible
• Validate critical predictions experimentally

Next steps after ADMET profiling

Compound optimization:

• Use results to guide medicinal chemistry efforts
• Focus optimization on compounds with acceptable ADMET profiles
• Balance potency improvements with PK optimization
• Generate decision trees for compound advancement

Experimental planning:

• Design in vitro assays based on prediction results
• Plan in vivo pharmacokinetic studies for promising compounds
• Develop formulation strategies for poorly soluble molecules
• Prepare safety pharmacology studies for high-risk compounds

Further analysis:

• Integrate with other prediction tools for comprehensive profiling
• Generate regulatory submission-ready ADMET data
• Create project reports with PK and safety summaries
• Feed results to quantitative modeling approaches

Research documentation:

• Maintain audit trails for all ADMET predictions
• Document analysis parameters for reproducibility
• Generate comprehensive reports for project archives
• Track prediction updates and model improvements

Read next