Virtual screening is a firmly established technique in computational drug design which has saved the pharmaceutical industry billions of dollars since its inception. The structure-based approach involves fitting potential small molecules to a protein target, to generate predicted ligand-protein interaction properties that can significantly reduce chemical search space. Ligand-based virtual screening does not involve the direct use of a 3D representation of the biological target but instead uses known active compounds as a reference structure for screening in chemical space. With a multitude of approaches available to the end-user, we discuss the technical and practical considerations for modern virtual screening to ensure projects are delivered efficiently, and at optimal cost, and the time to get a drug to market is as short as possible.
Hit identification in drug discovery presents an immense task to researchers, with current estimates putting total chemical space at over 1060 plausible compounds. The odds of finding active molecules against a particular biological target are, therefore, stacked against researchers, prompting the need for high-throughput screening stages in most projects. The monetary, material and time waste in lab-based screening is enormous, which has driven the widespread adoption of in silico techniques for early-stage hit identification.
Virtual Screening Increases Efficiency in Early-stage Drug Design
Virtual screening of compounds has several benefits when compared to lab-based approaches. Firstly, there are enormous gains in speed. Unlike lab-based approaches, virtual techniques do not require the purchase of chemical compounds and do not need to be physically loaded into laboratory apparatus prior to screening. A computational chemist only needs access to chemical structure databases, which are readily available in the modern era. There is also far less waste generated by computational approaches.
Most compounds purchased for lab-based screening do not yield a hit, thereby wasting the chemical material and time investment in running the experiment, as well as incurring unnecessary environmental impact. In virtual screening, however, waste is minimised, because the process can be heavily automated, and the only cost is computing time. In addition to speed and efficiency gains, in silico screening makes it possible to consider millions of compounds in a matter of hours, which is orders of magnitude higher than can be achieved in a lab-based screen. Virtual screening can, therefore, explore chemical space much more widely than lab-based wet screening setups, meaning that more hits can be identified in a given timeframe.
Outsourcing Virtual Screening Enables Access to Specialised Knowledge without Incurring Overheads
Virtual screening can have significant implications for the success of drug discovery projects, which is why expert knowledge is needed to maximise hit rates. The demand for this specialised expertise has led to the formation of Contract Research Organisations (CROs), focussed on providing outsourced computational modelling for early-stage drug design. The benefit of this kind of working model is that drug discovery organisations can very quickly access specialised knowledge centres at different stages of the project, without incurring the initial overheads of hiring experts internally. This kind of model has become popular and has driven the development of computational tools for collaboration between project stakeholders, specifically tailored to aid the collaborative Design-Make-Test-Analyse (DMTA) chemical design workflow.
In silico Virtual Screening is Increasingly Performed by Non-computational Chemists
Computational chemistry is often perceived as being extremely challenging because it involves mastery of concepts from both physical and computational sciences, which are often less familiar to lab scientists, introducing a barrier to their widespread use. Collaboration with CROs makes this specialist knowledge immediately accessible, there is a benefit in medicinal chemists adopting in silico techniques as part of their standard workflows. Recent years have, therefore, seen a focus on the development of scientific software which emphasises user-friendliness and the ability to be used ‘out-of-the-box’, meaning that virtual screening is becoming increasingly performed by non-computational chemists. Despite advances in the usability of computational modelling software packages, however, a good theoretical understanding of in silico methods remains essential in maximising the practical benefits of computational chemistry. As we discuss some of the key concepts from the structure and ligand-based drug design in the following paragraphs, it should be emphasised that neither structure nor ligand-based screening is superior; they are often regarded as orthogonal/complementary techniques, providing value in different scenarios.
