Accelerated molecular discovery

Combining physical, biological and data sciences
for insight and molecular discovery

Algorithmic designs to predict new classes of antimicrobials

Infections will be the deadliest diseases unless we act now. Up to 700,000 people die every year of drug-resistant infections, increasing to 10 million by 2050 and costing the economy a total of 100 trillion USD. To avoid “being cast back into the dark ages of medicine,” there is an urgent need backed by UK government strategy to conserve existing antibiotics and discover new ones.

Combining techniques at the interface between data, physical and biological sciences is a powerful tool to design new molecules and materials. Biomolecular simulation gives a detailed understanding of how molecules interact with targets, and data science can find patterns and design rules in large datasets. Together with biophysical experiments, these techniques create a feedback cycle that leads to quick insight and accelerated discovery.

In this context, we are exploring the rational design of new classes of antimicrobials. Our initial target is to mine protein sequences for short antimicrobial motifs that kill bacteria, but leave human cells unaffected. We seek to demonstrate how experimental and computer-generated datasets can be harvested to reveal features that inform new design concepts. We are developing new analysis and visualisation tools. We are revealing new molecular modes of action by which proteins interact and permeate cell membranes. The aim is to design new antimicrobial compounds unrelated to others.

Discovery loop

Prediction of permeability and conformational flexibility of peptide-based drugs

Current medicine can be divided into two categories: traditional small molecules and larger biologicals. Small-molecule drugs can be taken orally and are cheap to produce. However, they have a higher incidence of side effects. Biologicals are specific to their targets, but are more difficult to deliver — they cannot be administered orally and are metabolised quicker than small molecules.

Peptides fit between these two classes and may have the advantages of both classes — high target selectivity and potency with good membrane permeability and high metabolic stability. Hence, the peptide drug market is larger than $20 billion per year and growing. Novel classes of molecules have already produced blockbusters (>1 billion in sales). Biologics such as insulin have twice the success rate of small molecules in clinical trials. However, the industry is lacking tools to discover, characterise and optimise the properties of beyond-small-molecule therapeutics.

We are creating next-generation tools to characterise the properties of large molecules, focusing on permeability and conformational flexibility.

Next-generation materials modelling at the atomic and molecular scale

Despite steady advances in computational power, tradeoffs exist between completeness of the description of interactions and the time and length scales that are accessible for sampling. Thus, simulation of complex systems is achieved using highly simplified models, which may not capture the fundamental interactions responsible for emergent behavior. This is particularly acute in (but not unique to) bimolecular simulation where the pharmaceutical industry is seeking new innovations in primary methods.

We have developed a fundamentally new strategy for efficient, predictive models at the molecular scale with complete electronic responses, reduced reliance on empirical input from the condensed phase and the prospects for vastly improved transferability and predictive value. The resulting level of completeness in physical description enables isolated molecule properties to define model parameters, thereby eliminating fitting to condensed phase data. Thus, the framework provides a physical and intuitive basis for predictive, next-generation simulation.

Related blog post


[1] F. Cipcigan et al.
Structure and hydrogen bonding at the limits of liquid water stability
Scientific Reports 8(1718), 2018.

[2] Y.-W. Hsiao et al.
Cooperative Modes of Action of Antimicrobial Peptides Characterized with Atomistic Simulations: A Study on Cecropin B
J. Phys. Chem. B 122(22), 5908–5921, 2018.

[3] F. Cipcigan et al.
Accelerating molecular discovery through data and physical sciences: Applications to peptide-membrane interactions
J. Chem. Phys. 148, 241744, 2018.

M.-P. Pfeil et al.
Tuneable poration: host defense peptides as sequence probes for antimicrobial mechanisms
Scientific Reports 8(14926) 2018.

[5] A. Pyne et al.
Engineering monolayer poration for rapid exfoliation of microbial membranes
Chemical Science 8(2), 1105-1115, 2017.


[6] G. Henihan et al.
Label- and amplification-free electrochemical detection of bacterial ribosomal RNA
Biosensors and Bioelectronics 81, 487-494, 2016.

[7] F.S. Cipcigan et al.
Electronic coarse graining enhances the predictive power of molecular simulation allowing challenges in water physics to be addressed
J. Computational Physics 326, 222-233, 2016.

[8] V.P. Sokhan et al.
Molecular-scale remnants of the liquid-gas transition in supercritical polar fluids
Physical Review Letters 115(11), 117801, 2015.

[9] V.P. Sokhan et al.
Signature properties of water: Their molecular electronic origins
Proc. National Academy of Sciences 112(20), 6341-6346, 2015.

[10] A. Jones et al.
Electronically coarse-grained model for water
Phys. Rev. Lett. 110(22), 227801, 2013.


[11] L. Ryan et al.
Anti-antimicrobial Peptides. Folding-mediated Host Defense Antagonists
J. Biological Chemistry 288(28), 20162-20172, 2013.


[12] P.D. Rakowska et al.
Nanoscale imaging reveals laterally expanding antimicrobial pores in lipid bilayers
Proc. National Academy of Sciences 110(22), 8918-8923, 2013.

[13] S.N. Syed et al.
Cyclic denaturation and renaturation of double-stranded DNA by redox-state switching of DNA intercalators
J. American Chemical Society 135(14), 5399-5407, 2013.

[14] D.K. Corrigan et al.
Impedimetric detection of single-stranded PCR products derived from methicillin resistant Staphylococcus aureus (MRSA) isolates
Biosensors and Bioelectronics 34(1), 178-184, 2012.