Finding Superior Drug Candidates Through Advanced Computer-Aided Drug Design

Bringing a new drug to market is a costly endeavor with a constant risk of failure. Can high-performance computing and advanced computer-aided drug design identify superior drug candidates faster and at a lower cost than existing methods?

Early drug discovery centers around identifying hit compounds with a desired biological activity and optimizing these hits into leads for further development. But within this process, there are numerous challenges to overcome.

For example, the pool of available compounds may be too small to find good hits with the desired biological activity. Or, perhaps a drug candidate has pharmacokinetic or toxicity liabilities that need to be solved. Legal or intellectual property limitations may require the use of an alternative chemical scaffold. And sometimes, projects simply reach a dead end and must start over with completely new hits. 

“Our advanced and unique computer-aided drug design technology can solve these issues quickly and cost-effectively by identifying more diverse and more active drug candidates,” said Enric Gibert, CEO at Pharmacelera, a Barcelona-based company specialized in ligand- and structure-based virtual screening, in silico services, machine learning, and high-performance computing.

Traditional computational drug discovery

Enric Gibert, drug discovery, virtual screening, computer-aided drug design, pharmacelera
Enric Gibert, CEO at Pharmacelera

New methods using quantum mechanics-based calculations to describe molecules overcome these limitations and provide highly reliable virtual screening results. These advanced computer-aided drug design approaches hold great promise to transform drug discovery by supporting faster and less costly discovery and optimization of drug candidates, drug repurposing, and personalized medicine.

Computer-aided drug design and virtual screening have been used for many years to support early drug discovery. Both structure- and ligand-based virtual screening techniques have provided medicinal chemists and biologists with new compounds with a potential therapeutic activity. The downsides of these traditional computer-aided drug design methods stem from their mathematical simplifications and use of classical physics algorithms. 

Structure-based techniques require prior elucidation of the target’s crystal structure, which is not always available. Also, the scoring functions – the predictions of how well compounds bind to the target – tend to be oversimplified by the rigid conformation of the crystal structure and complexity of computing thermodynamic functions; this can result in false positive hits.

Standard ligand-based tools provide limited results because they usually only analyze the chemical structure of a known active compound and identify hits based only on the structural similarities.

Finding larger chemical diversity 

PharmScreen Workflow labiotech article 1
How does PharmScreen work?

Pharmacelera uses AI and machine learning to train and improve its models to mine an unexplored chemical space and find hits with a larger chemical diversity. The company has developed a software called PharmScreen, which uses a unique and superior 3D representation of molecules as an alternative and better ligand-based approach to traditional computational techniques. 

PharmScreen virtual screening software enables computational and medicinal chemists in drug discovery teams to explore a more structurally diverse chemical space and find new and biologically active hit candidates from libraries of millions of compounds. 

“Our software operates differently from other virtual screening tools on the market,” said Gibert. “PharmScreen uses quantum mechanics approaches to generate a 3D representation of molecules based on electrostatic, steric, and hydrophobic fields. The software then compares those fields and ranks the molecules in the screening library based on similarity to the reference.”

PharmScreen overcomes the bias commonly associated with other ligand-based approaches, because the compounds it identifies have different and innovative scaffolds but a similar pharmacological profile than the reference. It is also an excellent complement to structure-based approaches.

“Using quantum mechanics calculations, PharmScreen applies higher computing efforts than other methodologies in order to explore the chemical space in a more efficient and differential way,” Gibert explained.

Despite its computationally intensive nature, PharmScreen requires no special equipment, as the cloud-based software is run remotely from most standard computers. And for researchers lacking bioinformatics or computational chemistry expertise, Pharmacelera offers extensive support and in silico services to ensure a project’s success, from target identification to lead optimization.

Pharmacelera is enhancing its software capabilities to further improve its drug discovery support. Their newly developed application programming interface will allow PharmScreen users to customize and expand the software’s functionality. The company is also optimizing the tool so every screening campaign can assess bigger compound libraries in less time.

The results speak for themselves

In particularly challenging drug discovery areas such as central nervous system diseases or G protein-coupled receptors, PharmScreen overcomes the limitations common to standard structure- and ligand-based techniques.

A case study exploring a dataset of inhibitors of soluble epoxide hydrolase, a new therapeutic target for Alzheimer’s disease, shows that PharmScreen can efficiently discriminate real hits over decoys – compounds with similar structure to the reference but no activity – and find the most active compounds against the target in the top-ranked results.

Another case study on a dataset of 40 different G protein-coupled receptors, showed that using PharmScreen to calculate hydrophobic fields of compounds, allows the exploration of more diverse chemical structures. When compared to another commercial tool that bases its computations only on molecular shape, PharmScreen identified almost three times as many confirmed hits against the target of interest.

Custom in silico solutions for every client’s needs

In addition to its proprietary software, Pharmacelera offers tailor-made computational services to support the entire early drug discovery process: from designing compound libraries, hit identification and lead optimization using structure- and ligand-based methodologies, to discovering new molecular scaffolds for existing active compounds, understanding compound-target interactions, drug repositioning services, and more. 

The team’s experience working with varied clients allows them to consider different strategies and find the best approach for each client and project. 

“[The Pharmacelera team] worked closely with us to understand the problem and to provide a bespoke technical solution for our program in an iterative and flexible process,” reported Wren Therapeutics regarding their project with Pharmacelera.

And Macrophage Pharma was similarly pleased: “It was a very productive collaboration that allowed [us] to develop a structure-based drug design strategy to address the target of interest through iterative discussions and experimentation.” 

Ready to explore a greater chemical space? Find out more about PharmScreen and Pharmacelera’s in silico services! 


Images via Pharmacelera and digimolecules.com

The post Finding Superior Drug Candidates Through Advanced Computer-Aided Drug Design appeared first on Labiotech.eu.

Source

About the Author: Biotech Today

You might like