MUSE简单教程
#来源-muse tutorial or bookshelf
3.1 Quick Invention Experiment Tutorial
This tutorial shows how to use the Invention Editor in Muse to do a lead expansion experiment involving Estrogen Receptor (ER) agonists.
If you would prefer to watch a demo of this tutorial (approximately 4 minutes), please click here.
- Retrieve and save the file coumestrol.hits to your file system.
Do one of the following, depending on your platform:
• Windows: Right-click the link below and select Save Target As.
• Linux: Copy the contents displayed when you click the link below to a text editor (when finished, right-click in this frame and select Back to return to this page).
Save as a file called coumestrol.hits in your file system.
coumestrol.hits - In Muse, start a new Invention.
Click Invent New Structures in the Getting Started page, or select File > New > Invention Experiment from the menubar.
The first panel has a very brief description of the Invention process.
Click First Step. - Select and setup a scoring function.
For this example, the 3D pharmacophoric and shape similarity scoring function whose parameters have been set for lead expansion will be used.
From the Scoring Function pull-down, select Lead Expansion using Pharmacophore & Shape Similarity.
This scoring function is designed to focus the structure invention towards similar structures by preferring invented structures that are moderately high in similarity to the reference compound.
One reference structure, coumestrol, will be used in this example. Coumestrol is a natural estrogen receptor agonist.
Drag and drop the coumestrol.hits file from your file system into the grid. - Click Next Step.
- Specify any seed structures.
Seed structures are the parents of the first generation of compounds.
Since this is a lead expansion experiment, no seeds will be specified, which will allow Muse to generate the first generation of compounds.
Click Next Step. - Specify any preserved cores.
When substructures are specified as preserved cores, every invented structure will contain one of those core substructures. In this experiment, no cores will be preserved.
Click Next Step. - Specify invention settings.
For this example, ask to generate about 125 new structures.
From the first pull-down, select 125.
Click Next Step. - Specify any substructures to exclude.
Functional groups can be specified for exclusion from the invented structures. The Invention Editor has two exclusion lists that are available and selected for use by default. Additional groups can be specified for exclusion by creating a new exclusion list.
For this example, the default exclusion lists will be used. - Start the invention.
Click Invent.
As structures are invented they are presented in the Invention Results panel. This invention takes about 10-15 minutes. The Stop button in the Steps panel can be used at any time to end the invention and keep any intermediate results.
Each invented structure shows:
• Pharmacophore Similarity: The largest pharmacophore tuplet similarity found between the invented structure and each of the reference compounds.
• Shape Similarity: The largest steric tuplet similarity found between the invented structure and each of the reference compounds.
• Structural Similarity: The largest structural similarity, based on UNITY fingerprints, found between the invented structure and each of the reference compounds.
Note that compounds surviving through multiple generations will only show up once in the results. - Review the results.
Once the invention is complete, these additional columns are presented:
• First Generation: The generation in which the structure first appeared.
• Last Generation: The generation in which the structure last appeared.
Note that generation 0 contains the seed structures (in this case, provided by Muse).
Once the computation is complete, the results can be ranked using Pareto-ranking with the chosen scoring function.
Click Rank in the Steps panel.
After reading the information, proceed with by clicking Rank in the dialog.
A Rank column appears in the spreadsheet and the compounds are sorted from best to worst.
This concludes the Quick Invention Experiment Tutorial.
3.2 Lead Hopping Invention Tutorial
This tutorial shows how to use the Invention Editor in Muse to do a lead hopping study involving Estrogen Receptor (ER) agonists.
- Save the tutorial files to your file system.
There are several files that are needed for this tutorial. To save them to your file system:
Do one of the following, depending on your platform:
• Windows: Right-click a link below and select Save Target As.
• Linux: Copy the contents displayed when you click the link below to a text editor (when finished, right-click in this frame and select Back to return to this page).
Save as a file with the same name as the link (e.g., coumestrol.hits) in your file system.
Repeat for each of the other files.
• coumestrol.hits: The coumestrol structure that will be used as the reference compounds in this tutorial. If you ran the Quick Invention Experiment Tutorial, you already have downloaded this file.
• preserved_core.hits: A phenol and hydroxyl group that will be preserved in the invented structures.
• erag_actives.hits: Known ER actives. This file is not required to complete this tutorial. However, it has been made available to allow you to become familiar with the structures and for comparison against the invention results. - In Muse, start a new invention.
Click Invent New Structures in the Getting Started page, or select File > New > Invention Experiment from the menubar.
The first panel has a very brief description of the Invention process.
Click First Step. - Select and setup a scoring function.
The Invention process starts with specifying a scoring function to use and performing the setup for the chosen function. This tutorial involves using a 3D pharmacophoric and shape similarity scoring function whose parameters have been set to optimize lead hopping. This scoring function measures the 3D pharmacophoric and shape similarity of the invented structures against a set of reference compounds.
From the Scoring Function pull-down, select Lead Hopping using Pharmacophore & Shape Similarity.
This scoring function is designed to focus the structure invention towards novel structures by preferring invented structures that are lower in similarity to the reference compound.
One reference structure will be used in this example, coumestrol, which is a natural estrogen receptor agonist.
Drag and drop the previously saved coumestrol.hits file from your file system into the grid.
In the Scoring Function panel of the Invention Editor, click Next Step. - Specify any seed structures.
Specifying seed structures is an optional step in the invention process. Seed structures are a set of compounds that will be used as the parents of the first generation of compounds.
Seed structures and preserved cores are the two areas that can be used to direct the types of structures that are invented. Since the intention in this tutorial is to do lead hopping, no seeds will be specified and the program will be allowed to randomly generate the first generation of compounds. (Refer to Hints for Lead Hopping, Optimization, and Discovery for more details on when to use seed structures.)
Click Next Step. - Specify any preserved cores.
Specifying core substructures is an optional step in the invention process. Core substructures are either connected (single fragment) or disconnected (multi-fragment), and remain unchanged throughout the evolutionary process. Every invented structure will contain one of the core substructures.
When the core substructure is prepared for substructure searching, all implicit hydrogens are added, the atoms are typed, and then the Xs (or Rs) are stripped off. So implicit hydrogens are required, the Xs become empty valences, and the atom types prescribe the number of connections.
Preserved cores and seed structures are the two areas that can be used to direct the types of structures that are invented. Using small cores of R-groups or key fragments generates structures that include the groups you know are important. However, because the preserved part is small, the variation in the structures will hopefully result in some novel, lead hopping candidates. Refer to Hints for Lead Hopping, Optimization, and Discovery for more details on when to use preserved cores.
For this tutorial, a phenol and an hydroxyl group will be used as the preserved cores. The ER antagonists tend to have the aromatic rings but without the hydroxyl groups. So the aromatic ring(s) appears to be good for binding, and the hydroxyl groups seem to be important for being an agonist instead of an antagonist.
Drag and drop the previously saved preserved_core.hits file from your file system into the grid.
Notice that the Marked indicator is activated by default for the imported core. Only marked cores will be included in the invention.
In the Preserved Cores panel, click Next Step. - Specify invention settings.
The default settings will be used to generate about 500 new structures. For reproducibility, a specific random number seed will be entered.
Click Advanced Options.
Enter 12345 as the Random Number Seed.
Click Next Step. - Specify any substructures to exclude.
Functional groups can be specified for exclusion from the invented structures. The Invention Editor has two exclusion lists that are available and selected for use by default. Additional groups can be specified for exclusion by creating a new exclusion list.
For this tutorial, the default exclusion lists will be used. - Start the invention.
Click Invent.
As structures are invented they are presented in the Invention Results panel. This invention takes about 45 minutes. The Stop button in the Steps panel can be used at any time to end the invention and keep any intermediate results.
Although all compounds that were generated are reported, the total will be less than the number of generations multiplied by the number of compounds to generate because compounds that survive through multiple generations only show up once in the results.
Each invented structure shows:
• Pharmacophore Similarity: The largest pharmacophore tuplet similarity found between the invented structure and each of the reference compounds.
• Shape Similarity: The largest steric tuplet similarity found between the invented structure and each of the reference compounds.
• Structural Similarity: The largest structural similarity, based on UNITY fingerprints, found between the invented structure and each of the reference compounds.
Once the invention is complete, these additional columns are presented:
• First Generation: The generation in which the structure first appeared.
• Last Generation: The generation in which the structure last appeared.
Note that the 0 generation contains the seed structures (if none were provided, then Muse generates a random initial set for you).
Once the computation is complete, the results can be ranked using Pareto-ranking with the chosen scoring function.
Click Rank in the Steps panel.
A Rank column appears once the computation is complete and the compounds are reordered by rank from best to worst.
Scanning through the results, a phenyl hydroxyindole-like core is frequently seen in structures that were produced in later generations.
Different from the coumestrol, structures with this core can be submitted to Muse as seeds for another invention to explore the possibilities along this core type.
This concludes the Lead Hopping Invention Tutorial.