Computational Chemistry is a branch of chemistry that uses mathematical calculations and computer simulation to solve chemical problems. In drug discovery Computational Chemistry helps to find answers to the following questions:
- What molecules should I synthesize next?
- What structures have largest chance of success for binding?
- How can the compound be optimized to reduce side-effects?
- Have I missed any potential hits from the dataset?
Wildcard Pharmaceutical Consulting has experience in the area of computational chemistry that is related to drug discovery and has a large overlap to computational biology and bio-structural modeling. If you have asked questions like the ones above, you are welcome to contact us for a free and confidential discussion. With insight into your situation, we can better suggest how computational chemistry can be used successfully.
Benefits of computational chemistry
- Save Money and resources by evaluating potential bioactivity and avoiding synthesis of hopeless ligands.
- Save time in hit-to-lead optimization by faster feedback cycles and better design for new compounds.
- Lower the risk of project failure by early assesment of possible off-target activity or bad ADME/tox properties.
There is no single computational chemistry technique which fits all projects. The selection of the most appropriate and useful one is dependent on the available information and the stage of the project. In the early phases, the available information is very limited, whereas later it increases. This can be in a form of bioactivity data for large ligand series or biostructural information about the binding site. External information from scientific literature and patents are also applicable.
Even with scarce information computational methods can be used to enhance the chance of success. As more information gets available more opportunities for applying computational methods appear and so does the need for effective data management. Because of our experience in lab data management, chemoinformatics, chemical databases and Linux servers we can in addition to the computational chemistry consulting, support projects with information management and thus maximize the utilization of the available information.
In order to solve client’s specific needs, Wildcard Pharmaceutical Consulting is using multiple open source software and libraries together with custom python programs. Additionally third party tools and software packages (e.g Schrödinger) can be utilized.
Scenarios for Computational Chemistry Consulting
Overall the computational chemistry methods in drug discovery can be divided into two areas, Ligand based approaches and Structure based
approaches. It depends if information about the ligands or the target structures are used as basis of the method or not.
Ligand based approaches
The ligand based computational chemistry techniques are using the information and knowledge about active ligands. The methods are used to optimize, search or evaluate novel compounds before any resources are spent on procuring or synthesizing them. Depending on the available amount of previous information different methods can provide value. Even with no examples of bio active ligands, it is still possible to design focused libraries. Using focused libraries in high throughput screening will enlarge the hit rate and chance of finding a good lead for subsequent optimization. It is beneficial to additionally filter the ligands in the test library for possible adverse effects such as off-target activity, toxicity and bad ADME properties. This way resources are saved by focusing on promising rather than problematic ligands.
When the number of known active ligands grows, the possibilities for more advanced analysis of structure activity relations (SAR) and pharmacophore modelling improve. In SAR analysis the structural features and chemical properties are linked to the wished activity. The understanding of the connection between chemical structure and the wanted activity can be used in the subsequent optimization of the ligand series. Pharmacophore modelling builds a model of the ligand features that are needed for pharmacological action. The molecules are reduced to a selection of molecular features and their spatial relationship. The pharmacophore model can be used to screen databases for molecules that also fit the model. In this way completely new scaffold can be found, which would not be found by similarity searching in the database.
In QSAR the SAR relationship is modeled and analyzed with statistical and machine learning techniques. The model can be a linear model such as multiple linear regression or partial least squares (PLS). More advanced nonlinear models such as support vector machines (SVM), random forests or neural networks can be used if the data volume and quality are supporting it. Depending on the project the prediction can be a classification (toxic/non-toxic) or a continuous property such as ligand affinity.
Methods such as 3D-QSAR can be used to get a better understanding of the binding mode and most likely bio active conformation. Libraries can be tested for fitting to the model. The 3D-QSAR model can also give feedback about which areas are available for further expanding and which areas are less essential for binding and thus available for optimization to other properties such as solubility.
Wildcard Pharmaceutical Consulting can provide ligand based computational chemistry consulting services that are customized and adapted to your specific project.
Structure Based Approaches and Ligand Docking
Experimental techniques such as X-ray crystallography or Protein NMR provide information about the structure of the target binding site at atomic level. If no information for the specific target in question is available, homology modelling can sometimes be used as a substitute to get early information about the structure of the drug target. With Homology modelling structures from close homologs are used as templates to build a model of the wanted target. The technique is thus useful for generating structural models of drug targets with an unknown structure. It is possible to develop good homology models for protein families with over 40% amino acid identity. With special care reasonable models can be produced with down to 20% sequence identity between the known structure and the target.
Having structural insight is no turn-key solution due to the inherent receptor flexibility, but structural data can nevertheless play a crucial role in speeding up the discovery of novel ligands and support efficient selection of compound candidates for testing.
The receptor and binding site flexibility can be examined with molecular dynamics simulations, which are also useful for validation and refinement of homology models. Knowledge about the flexibility of a target receptor is useful as it can show which areas of the receptor binding site are dynamic and which are more rigid. The dynamic areas can to a larger degree accommodate steric overlap with proposed ligands, whereas the rigid part must have a more perfect complementary fit with the ligand structure. If a ligand is bound to the experimentally determined structure of the target, the interactions between the ligand and the receptor can be analyzed. The analysis can guide modifications to the ligand to either increase affinity or change key structural features.
Automated ligand docking is the process of predicting the binding mode and affinity for a given ligand to the receptor. This is done with specialized programs that evaluate and optimize the ligand position and conformation with respect to the receptor. They do so by selecting the most likely binding mode via a scoring function. The technique is useful for docking known hits with unknown binding mode. The binding mode can be analyzed and the results used to provide structural feedback for design and optimization of the next round of compounds in hit-to-lead medicinal chemistry optimization. On the blog is some tips and caveats about some open source docking programs
Docking can also be used for virtual high throughput screening (vHTS) to design target focused libraries for subsequent procurement and testing. By using the structural information, the screening campaign for novel hits can be changed from a chance based lottery into a knowledge based approach with a much higher chance of success.
Alternatively the structural information about the binding site can be used to de-novo design ligands to fit the protein receptor site. This must be done in close collaboration with medicinal chemists to ensure synthetic feasibility of the proposed ligands. Fragment based approaches are also a possibility, where molecular fragments are docked to the binding site and later joined.
Wildcard pharmaceutical Consulting can provide computational chemistry consulting services in the form of homology modelling to provide structural information for structure based drug design and also do automated docking. Contact us for a free and confidential meeting where we can discuss your specific situation and advise about the possibilities for using computational chemistry consulting as part of your drug discovery project!
Selection and Validation of Computational Chemistry Methods and Software
Depending on your specific research project and needs, the computational chemistry consulting will often involve a mix of techniques and methods. We are able to combine and select methods and software that fits the needs of your project. We always strive to use the best solutions as well as validate and benchmark our selection of methods to the specific target and project.
If you plan to, or already have purchased, computational chemistry software for in-house use, check our IT services page. We offer help with selection, setup, installation and maintenance of servers and scientific software.
You are always welcome to contact for an informal but confidential chat about how computational chemistry can solve your needs.