Iterative Design page

Orion’s Iterative Design page incorporates functionality that allows users to generate new analogs of existing compounds. The Iterative Design page is organized as a 3-stage process: (1) the selection of originating matter, (2) the application of generative methods, and (3) the analysis and filtering of generated structures. At stage 3, structures can be sent back to stage 1 and used as further originating matter, iterating the process. Switching to the Iterative Design page automatically populates stage 1 with either selected structures or the current active dataset. Any key structures are also made available. Features from the Analyze page are borrowed to help select structures. Generation methods in stage 2 include 2D and 3D editing and several innovative chemical modification approaches. Structures created by any method are included in the stage 2 pool and can act as originating material for other methods or can be deleted at any point. Those retained are passed to stage 3. In this stage, generated structures can be compared to the originating matter (e.g., bar graphs of the basic properties of both). Structures can be added back to stage 1 or saved as a new dataset.

_images/iterativedesign.png
Iterative Design Page Layout

Name

Description

Design Session

Allows users to move through the Design stages 1 - 3, and back to stage 1.

Scatter Plot

Allows users to choose compounds for the next Design stage. By default, the plot shows all compounds the user originally selected in the Analyze page.

Spreadsheet

Allows users to choose compounds for the next Design stage. By default, the spreadsheet displays all compounds the user originally selected in the Analyze page.

Hint

Molecules for the Design cycle can be selected from either the scatter plot or spreadsheet.

3D Editor

The 3D editor allows users to modify a structure in 3D and has the same molecular editing capabilities as the 3D editor in the 3D Modeling page. After molecules to modify are selected in Design stage 1, all molecules are propagated to the 3D editor.

_images/iterativedesign3d.png

Highlighting a molecule from the list activates the editing tools. After making some modifications to a molecule, the edited molecule can be saved as a new molecule or analog. The new molecule is then placed in a container for the Evaluate stage.

2D Editor

The 2D editor allows users to modify a structure in 2D and has the same molecular editing capabilities as the 2D editor in the Analyze page. After molecules to modify are selected in Design stage 1, all molecules are propagated to the 2D editor.

_images/iterativedesign2d.png

Highlighting a molecule from the list activates the editing tools. After making modifications to a molecule, the edited molecule can be saved as a new molecule or analog. The new molecule is then placed in a container for the Evaluate stage.

Matched Molecular Pairs

The Matched Molecular Pairs (MMP) subpage is prepopulated with the molecules selected in Design stage 1 and, optionally, analogs saved from the other design methods. MMP uses the algorithm described in the OpenEye OEMedChem TK.

_images/iterativedesignmmp.png

Users can select one of the compounds in the prepopulated list and generate up to 200 analogs. Analogs are tagged with the following molecular properties: molecular weight, the number of hydrogen bond donors, the number of hydrogen bond acceptors, and XLogP. Selected analogs accumulate in the molecule list on the left of the subpage. All new analogs of interest can be further evaluated and selected in the Evaluate stage.

Hint

Remove unwanted analogs by clicking the (X) icon next to the analog.

Near Fragment Replacement

The Near Fragment Replacement (NPR) subpage is prepopulated with molecules selected in Design stage 1 and, optionally, analogs saved from the other design methods. The NPR algorithm uses molecular fragments from the GSK TCAMS1 Malaria Dataset. This dataset was mined to identify a set of fragments that are commonly used in medicinal chemistry. The input molecule is broken into fragments, and a series of new fragments that are close in graph-edit space are identified. Each original fragment is replaced in succession by each of the nearby fragments generating, analogs of the original molecule. The fragments are replaced one at a time to keep each new analog about 60-80% identical to the original molecule.

_images/iterativedesignfr.png

Users can select one of the compounds in the prepopulated list and generate upto 200 analogs. Analogs are be tagged with the following molecular properties: molecular weight, the number of hydrogen bond donors, the number of hydrogen bond acceptors, and XLogP. Selected analogs accumulate in the molecule list on the left of the subpage. All new analogs of interest can be further evaluated and selected in the Evaluate stage.

Hint

Remove unwanted analogs by clicking on the (X) icon next to the analog.

Reagent Join

The Reagent Join (RJ) subpage is prepopulated with molecules selected in Step 1 of the Design process and, optionally, analogs saved from the other design methods. The RJ algorithm allows users to select a specific set of reagents. Users can select a portion of the original molecule that defines where the reagents will be joined to the original molecule. While introducing specific reagent types, this method does not require specific chemistry to be present but instead relies on users to determine appropriate connections for their application. This method is similar to, but more constrained than the Explore Side Chains method.

_images/iterativedesignrj.png

Users can select one of the compounds in the prepopulated list and generate up to 200 analogs. Analogs are tagged with the following molecular properties: molecular weight, the number of hydrogen bond donors, the number of hydrogen bond acceptors, and XLogP. Selected analogs accumulate in the molecule list on the left of the subpage. All new analogs of interest can be further evaluated and selected in the Evaluate stage.

Hint

Remove unwanted analogs by clicking the (X) icon next to the analog.

Explore Sidechains

The Explore Sidechains (ES) subpage is prepopulated with molecules selected in Step 1 of the Design process and, optionally, analogs saved from the other design methods. The ES algorithm uses a shape and chemical diverse set of side chains that have been identified from an initial collection of available reagents. These side chains can be used to replace a select portion of a molecule. If the portion selected for replacement is small, this method grows a new side chain onto the molecule. If the portion selected for replacement is larger, this method replaces or possibly shrinks the original molecule. This method explores new chemical matter without any notion of reagent selection or synthetic accessibility. It is similar to but less formal and specific than the Join Reagents method.

_images/iterativedesignes.png

Users can select one of the compounds in the prepopulated list and generate up to 200 analogs. Analogs are be tagged with the following molecular properties: molecular weight, the number of hydrogen bond donors, the number of hydrogen bond acceptors, and XLogP. Selected analogsaccumulate in the molecule list on the left of the subpage. All new analogs of interest can be further evaluated and selected in the Evaluate stage

Hint

Remove unwanted analogs by clicking the (X) icon next to the analog.

Evaluation

The Evaluation subpage comprises a regression model bar graph, 4 molecular properties bar graphs, and a spreadsheet of the original molecules and analogs. The bar graphs allow users to compare the properties of the newly designed analogs (in yellow) to the original molecules (in blue). The spreadsheet displays all of the molecules (original and analogs) and their molecular properties. Regression models previously computed in the Analyze page, if present, are displayed in the Model Plot.

_images/iterativedesignevaluate.png

Newly designed analogs can be submitted to a subsequent design iteration or saved as a new dataset.