Computer-Aided Drug Design (CADD)
- ADMET Modeling and Prediction
- De Novo Drug Design
- Ligand Based Virtual Screening
- Quantum Mechanics for Target Selection
- Structure-Based Virtual Screening
- DNA-Encoded Library Technology (DELT)
- Fragment-Based Screening
- High Content Screening (HCS)
High Throughput Screening (HTS)
- Automated HTS Platform
- Biochemical assays in Hit Characterization
Biophysical Assays in Hit Characterization
- BLI for Affinity-based Hit Screening
- CD Spectrometry for Protein Structure Determination
- ITC for Binding Assessment
- MS for Structure Confirmation
- MT for Binding Affinity Measurement
- NMR Spectrometry for Tareget identification and Characterization
- SPR Spectrometrys for Structure Determination
- TSA for Protein's Stability Evaluation
- Cellular assays in Hit Characterization
- Drug Repurposing
- Hit Screening
- HTS Assay Development
- HTS Compounds Libraries
- HTS Data Management
- Virtual Screening (VS)
Drug Discovery Services
- Experienced and qualified scientists functioning as project managers or study director
- Independent quality unit assuring regulatory compliance
- Methods validated per ICH GLP/GMP guidelines
- Rigorous sample tracking and handling procedures to prevent mistakes
- Controlled laboratory environment to prevent a whole new level of success
SVM for Lead DiscoveryINQUIRY
Support Vector Machine (SVM) has been explored as a ligand-based virtual screening (VS) tool for facilitating lead discovery. The goal of this method is to create a decision boundary that separates two classes of points according to the active and inactive compounds. A key characteristic of SVM is to reduce the error on training data, minimize the complexity of models and avoid the overfitting with the application of the structural risk minimization approach.
Fig.1 Illustration of training a SVM virtual screening model and using it for searching inhibitors of an individual target. (Xiao, H. M.; et al. 2010)
Advantages of Support Vector Machine
- Easy to understand: Simple geometric interpretation.
- Nonlinear decisions based on the use of kernels.
Application in Drug Discovery
- SVM is applied to predict the distinction between active compounds and inactive ones through binary class labeling.
- SVM can also be utilized to rank compounds in database according to their activity potential in virtual screening.
- SVM can assess potency of candidates and theirs druggability.
- We have established an advanced SVM platform to optimize the screening process using the maximum margin hyperplanes.
- Our SVM technologies can separate the active from the inactive compounds rapidly and have the largest possible distance from any labeled compound.
- We have experience in choosing right SVM parameters to develop SVM models.
- Our teams apply the cross validation-derived statistic to select one among the different possible models.
- We provide both single and multi-target prediction with high success rate.
- At BOC Sciences, our professional scientists are capable of performing data processing and interpretation.
- Xiao, H. M.; et al. In-Silico Approaches to Multi-target Drug Discovery. Pharm Res. 2010, 27(5): 739-749.
※ It should be noted that our service is only used for research.