Can an automated workflow reproduce a plausible Imatinib–ABL1 docking setup?
The session assembled a computational workflow around Imatinib and UniProt P00519. It is a workflow demonstration using an ABL1 AlphaFold prediction—not a fusion-specific BCR–ABL experimental complex.
- Ligand
- Imatinib, C29H31N7O, molecular weight 493.6
- Structure
- AlphaFold v6 model for ABL1, UniProt P00519
- Selected site
- fpocket #29, druggability 0.540, volume 1,064.8 ų
- Docking
- Vina 1.1.2, 25 Å cube, exhaustiveness 32
From compound record to ranked docking poses
The workflow produced a favorable score, with important structural qualifications
Ten poses ranged from −8.8 to −9.4 kcal/mol in the Vina scoring function.
The selected pocket overlapped 12 of 39 known contact residues cited in the report.
Four of five drug-likeness filters passed; the Ghose filter failed on molecular weight.
The report flags low solubility, moderate CYP/hERG risk, and QED 0.389.
A docking score is not a measured affinity or confirmation of potency
- The AlphaFold model represents ABL1, not the BCR–ABL fusion or an experimental Imatinib-bound structure.
- The predicted ground-state conformation does not capture the DFG-out state associated with Imatinib binding.
- Rigid-receptor docking omits induced fit, explicit solvent, entropy, and cellular context.
- Vina’s score is a ranking heuristic, not a validated binding free energy.
- The ADMET values are rule-based estimates and require experimental or validated predictive follow-up.
Useful as a transparent workflow demonstration and pose-generation exercise; insufficient to confirm binding strength, selectivity, potency, or clinical behavior.
Reproducibility artifacts recorded in the session
imatinib_pubchem.csvCompound propertiesimatinib_admet_summary.csvRule-based property screenAF-P00519-F1.pdbAlphaFold ABL1 modelfpocket_results.csvPocket rankingsselected_pocket.jsonSelected-site metadataposes_all.sdfDocked poses