Train New Model
Last updated
Was this helpful?
Last updated
Was this helpful?
To generate molecules through Hyper Screening X, the model must first be trained.
Click the [Train new model] button on the model management screen to launch the model training interface.
Training a new model requires setting up two stages:
First, select the protein that will serve as the basis for the model’s training.
Only proteins of the ‘3D structure’ type currently added to the project can be selected. After selecting the protein, click the [Next] button to move to the next step.
(*Only one protein can be selected.)
If no proteins have been added, click the [Go to Protein management] button at the top to go to the Protein management page and add a protein to the project first.
After selecting the protein, you need to enter the basic information required for model training and set the candidate molecule conditions for molecule generation by the model.
(1) Enter Basic Model Information
Model name
Enter the desired name for the model.
The model name cannot be duplicated within the project.
Model tag
The tag that will be added to the ‘Hyper Screening X model’ field of the molecules generated by the model.
Tags cannot be duplicated within the project.
Molecule code
The code that will be used when naming molecules generated by this model.
The name of the generated molecule will be displayed as ‘Molecule code-Generation number’.
The molecule code can only be entered up to 10 characters without spaces in English.
Model description
Freely enter a description of the model.
(2) Set Physicochemical Property Conditions
These are the physicochemical property conditions for candidate molecules to be applied when adding the model.
[Conditions Setting Precautions]
The range of each property condition can vary depending on whether the Lipinski rule is On or Off.
LogP
2~7
2~5
Molecular weight
200~800
200~500
TPSA
0~120
0~120
Each property condition has a minimum range that can be set. (If the range is set too narrowly, the number of generated molecules may be small, or molecules with poor binding scores may be generated.)
LogP
The difference between the set maximum and minimum values must be at least 2.
Molecular weight
The difference between the set maximum and minimum values must be at least 300.
TPSA
The difference between the set maximum and minimum values must be at least 50.
Once the basic information for the model and the physicochemical property conditions of the candidate molecules are set, click the [Start training] button to start the model training.
Click the [Start training] button on the last confirmation screen before starting the model training, and the model training will begin.
Once model training starts, the status of the model training changes to ‘In Progress’ and will be completed within a maximum of 48 hours.
