ICM-VLS

ICM-Pro Virtual Ligand Screening (VLS) Module, developed by MolSoft L.L.C., is a powerful add-on to ICM-Pro designed for structure-based and ligand-based virtual screening in drug discovery. It combines genuine internal coordinate docking methodology with a sophisticated global optimization scheme, delivering fast and accurate flexible-ligand docking and scoring. The module is aimed at computational chemists, medicinal chemists, and drug discovery teams in pharmaceutical, biotech, and academic settings who need to screen large compound libraries against biological targets to identify lead candidates. The average yield of lead candidates is as high as 10%, and in many cases only the top 1% of hits need to be tested experimentally to find a lead. This product also includes all chemistry functionalities found in ICM-Chemist and ICM-Chemist-Pro.

MolSoft's ICM-VLS has demonstrated industry-leading performance, ranking first place in the D3R Grand Challenge 2 and 3 (2017 and 2018) for both docking pose prediction and affinity prediction. It also ranked first in an independent 2018 evaluation of six covalent docking methods, outperforming all other programs in reproducing experimental binding modes and scoring. Notably, Novartis scientists used ICM-VLS in the world's largest virtual screen, distributing jobs across more than 30,000 cloud computing cores to complete 10.6 years of drug compound computations in just 9 hours, resulting in three new lead compounds.

Core Docking and Screening Features

• Fast and accurate flexible-ligand docking and scoring procedure using internal coordinate methodology with global optimization.
• Screen large databases from chemical vendors (SDF files) or from target-specific libraries, including MolCart Compounds with over 9 million chemicals, ZINC, and eMolecules.
• Index large SDF files for efficient high-throughput virtual screening workflows.
• Automatically convert 2D database compounds to 3D, add hydrogens, and assign charges, including SMILES to 3D and 3D to SMILES conversions.
• Multiple methods for incorporating receptor-induced fit and receptor flexibility.
• Template docking for pharmacophore-based drug design — match templates by atom name, substructure, or 3D pharmacophore APF.
• Dock on-the-fly Markush-generated libraries and focused libraries.
• Apply database filters for criteria such as molecular size, number of hydrogen bond donors/acceptors, and number of torsions.
• Scan substituted derivatives of lead compounds.
• Covalent docking and screening using reaction-based approaches.
• Fragment docking, screening, and linking capabilities.
• Use the graphical user interface (GUI) to set up all stages of screening, or run from the command line in batch mode.

Ligand-Based Screening: Atomic Property Fields (APF)

• Ligand-based virtual screening using the Atomic Property Field (APF) method developed by MolSoft (Totrov 2008).
• APF is a 3D pharmacophoric potential implemented on a continuously distributed grid, usable for both ligand docking and scoring.
• APF can be generated from one or more high-affinity scaffolds, with seven empirically derived physico-chemical properties assigned: hydrogen bond donors, acceptors, Sp2 hybridization, lipophilicity, size, electropositive/negative character, and charge.
• A single ligand atom can contribute to multiple property fields; multiple spatially consistent ligand atoms produce a strong pharmacophore signal.
• APF has been extended to support multiple flexible ligand alignments using an iterative procedure.
• APF uses Monte Carlo minimization in atomic property field potentials in conjunction with standard force-field energies.

Results Analysis Features

• Easy-to-use graphical user interface for post-screening analysis.
• One-click hit list generation with rapid browsing of ligand binding poses.
• Interactive browsing of scan results with diverse display features including hydrogen bond and surface contact views.
• Filter results by charge, buried surface area, volume change, contacts, distance, and selection overlap.
• Analyze scan performance using score and property histograms and plots.
• Cluster docking results by substructure and APF similarity to obtain a diverse set of compounds for experimental testing.

Validated Success Stories

• Identification of lead compounds across a wide range of therapeutic targets including iNOS, neuropilins, serotonin transporters, dopamine receptors, LSD1, FtsZ, PERK, AKR1B10, EGFR, and many others.
• Screening campaigns ranging from tens of thousands to over 1.2 million compounds, with experimentally confirmed inhibitors and antagonists reported in peer-reviewed publications.
• First demonstration that GPCR homology models can be used for antagonist discovery by virtual screening.
• Successful identification of covalent inhibitors, fragment-based leads, allosteric modulators, and natural product-like compounds.
• Applications spanning oncology, infectious disease, neurology, and other therapeutic areas.

ICM-VLS supports both GUI-driven and command-line batch workflows, making it suitable for individual researchers and large-scale cloud-based deployments. Its cloud scalability has been demonstrated at industrial scale with Novartis, and the platform integrates seamlessly with MolSoft's broader ICM-Pro ecosystem for structure preparation, receptor modeling, and lead optimization.