OEToolkits 2024.1¶
Release Highlights¶
Release Highlights 2024.1.0¶
EON: Overlay Optimization with Shape and Charge Density¶
EON, the tool for molecular similarity based on shape and electrostatics, has been reimagined with the addition of charge density as a descriptor for electrostatic properties. By replacing electrostatic potential with molecular charge density as the primary descriptor of electrostatic properties, EON no longer requires the input molecules to be pre-aligned with the query molecule. Instead, EON now aligns the input molecules using shape- and electrostatics-based overlay optimization, where electrostatics are modeled via molecular charge density. EON still has the option to use electrostatic potential as a measure of similarity, if desired. Figure 1 shows an example of EON overlay with charge density visualization.
Figure 1. EON shape charge density overlay between two molecules.¶
Even with the addition of overlay optimization within the workflow, the new EON is now significantly faster compared to the previous version; and yet, as a tool for virtual screening with shape and charge similarity, the behavior of EON remains the same. When compared to ROCS, EON lags slightly when measured in terms of virtual screening success rates but excels in finding hits that are more chemically diverse.
Performance of the previous and current versions of EON, as compared to ROCS,
are displayed in Figure 2. The comparisons were performed using multi-query directory of
useful decoys (MDUD) datasets containing 38 targets. The current version of EON
overlays and scores with shape and charge similarity. A command line flag -potential supports the rescoring
of shape- and charge-aligned conformers with shape and electrostatic potential similarity. The current default version
of EON is now almost as fast as ROCS, as can be seen in the right-hand plot below.
Figure 2. Comparison of performance of EON and ROCS for virtual screening success rate (left); for chemical diversity of actives found (middle); and for the speed of calculations (right). “Current” refers to the default version using shape and charge density for similarity, and “current (potential)” refers to EON using shape and electrostatics.¶
Bioisostere TK: New Toolkit for Bioisosteric Fragment Replacement¶
The 2024.1 release introduces Bioisostere TK, a new toolkit for bioisosteric analog generation. It involves replacing a portion of a lead compound with fragments that have similar shape and electrostatics but with potentially novel connectivity and chemistry. This provides toolkit level access to the existing BROOD application functionality.
Bioisostere TK also brings several additional functionalities. At its core are tools for fragment-based modeling, such as fragment creation, fragment conformer generation, and fragment alignment and similarity in 3-dimensional space. Bioisostere TK brings OpenEye’s shape-, color-, and electrostatics-based molecular similarity tools into fragment-based similarity. It also exposes tools for the generation of 3D molecule conformers with fragment replacement.
An example of fragment alignment with Bioisostere TK based on shape and color is shown in Figure 3. The attachment points in both the reference and fit fragments are marked in purple. The fragment overlay highlights how special attention is paid to the overlay of the attachment points.
Figure 3. Fragment overlay between a reference and a fit fragment.¶
FLYNN: Ligand fitting for Cryo-EM¶
FLYNN, the tool for fitting small molecules to electron density, has been extended to support density maps generated through cryo-electron microscopy (cryo-EM).
FLYNN is a powerful tool for ligand fitting that uses adiabatic fitting to generate the best-fitting, lowest strain conformers consistent with experimental electron densities and structural models. This tool was one of the first to enable high-quality, physics-informed pockets and poses from X-ray crystallographic models. FLYNN has now been updated to enable researchers to access the same powerful insights from maps and models generated through cryo-electron microscopy (cryo-EM).
Figure 4. Ligand fitting with FLYNN to predict binding pockets and poses consistent with electron density from cryo-electron microscopy (cryo-EM).¶
Hydrogen Placement Updates in SPRUCE¶
In the 2024.1 release of SPRUCE, several improvements have been made to OEPlaceHydrogens to handle metal cation chelation effects on the nearby protonation states, specifically for ligands. The changes shown in Figure 5 below highlight new state and optimized geometries for sulfonamide, hydroxamate, phosphate, and borate groups. In addition, changes have been made around geometries for sp3 secondary amines. Lastly, improvements have been added in OEProtonateDesignUnit to allow for protonation state changes related to the detection of unfavorable acceptor–acceptor clashes, resulting in a more favorable hydrogen bond network.
Figure 5. Examples of improved chelator interactions and deprotonation rules for different moieties: hydroxamate, sulfonamide, and phosphates/borates. Also, on the bottom right, an example of improved interaction networks with acceptor–acceptor heuristic fixes.¶
Supported Platforms¶
Package |
Versions |
Linux |
Windows |
macOS (x64 and arm64) |
Python |
3.9 - 3.12 |
RHEL8/9, Ubuntu20/20-ARM/22 |
Win10/11 |
12, 13, 14 |
C++ |
RHEL8/9, Ubuntu20/20-ARM/22 |
Win10/11 (VS2022) |
12, 13, 14 |
|
Java |
8, 11, 20 |
RHEL8/9, Ubuntu20/20-ARM/22 |
Win10/11 |
12, 13, 14 (only Java 20 support for arm64) |
C# |
Win10/11 (VS2022) |
General Notices¶
Support for Java 20 is now available on all platforms.
Support for macOS 14 has been added.
Python 3.12 is now supported on all platforms.
This release will be the last to support Ubuntu 20.04. Support for Ubuntu 24.04 will be added in the next release.