The virtual screening method has now become an indispensable and important technology in the development of new drugs. The traditional virtual screening method refers to the use of computer algorithm programs based on physical chemistry for a given target protein to rank the small molecule compounds in the database, and select the potential ligand molecules of the target protein from them, and then pass the experiment to be verified. In recent years, another virtual screening method has gradually gained the favor of researchers. This method is just the opposite of the traditional virtual screening method. For a given ligand molecule, the target protein that can bind to it needs to be selected from the database, so it is called the inverse virtual screening (IVS) method.
At present, many different types of IVS methods have been developed. According to the different methods, they can be roughly divided into the following categories: ligand-based methods, binding site comparisons, protein-ligand interactions fingerprint analysis (protein-ligand interaction fingerprints) and docking-based methods. We perform inverse screening through shape similarity screening, pharmacophore model screening or inverse protein-ligand docking to find unknown targets, unexpected targets or secondary targets for small molecule drugs. These three different calculation methods are complementary and can be combined with each other. In contrast, shape and pharmacophore screening is simpler and faster, while inverse docking is more complicated and slower.
Figure 1. The principle and workflow of shape screening, pharmacophore screening, and inverse docking.( Hongbin H,et al.2018)
In the absence of reference compounds, it takes about 25 years to design and develop new active molecules from scratch. Due to the development of the field of artificial intelligence, there have been some new developments in the design of compounds from scratch. We provide an interesting method for inverse virtual screening, which is a variational autoencoder, which consists of two neural networks, an encoder network and a decoder network. The encoder network converts the chemical structure defined by the SMILES representation into a real-valued continuous vector as the latent space. The decoder can convert vectors from this latent space into chemical structures. This feature is used to find the optimal solution for the latent space, and inverse these vectors into real molecular structures through the decoding network. For most decompilations, one molecule is dominant, but there is less possibility of minor structural modifications. The latent space representation is used to train a model based on QED drug similarity score and synthetic accessibility score SAS. A molecular path with improved target properties can be obtained.
|Project name||Inverse Virtual Screening|
MedAI can provide you with the following service but not limited to:
|Cycle||Depends on the time you need to simulate and the time required for the system to reach equilibrium.|
|Product delivery mode||The simulation results provide you with the raw data and analysis results of molecular dynamics.|
The IVS method is favored by new drug developers mainly because it has important application prospects in the following aspects: drug target identification, drug repositioning, and drug side effects/toxicology (side effects /toxicity) research. The identification of drug targets helps to understand the mechanism of drug action, and then effectively improve the drug to enhance its efficacy. New use of old drugs is to find new target proteins for existing drugs to achieve the purpose of treating new diseases. Since all the compounds that have been formulated have passed the screening of side effects and clinical trials, this strategy can greatly shorten the time and cost of drug development. In terms of drug side effects/toxicology research, if the drug’s side effects can be effectively predicted in the early stage of new drug development, that is, to determine other proteins that can bind to the drug in addition to the target protein, it will greatly improve the drug’s passing rate in later clinical trials.
Please submit a detailed description of your project. We will provide you with a customized project plan to meet your research requests. You can also send emails directly to firstname.lastname@example.org for inquiries.
Tech Trends in Artificial Intelligence What's up, what's down, what lies ahead, only found in MedAI!