Predicting how molecules will react is vital for the discovery and manufacture of new pharmaceuticals, but historically this has been a trial-and-error process, and the reactions often fail. To predict how molecules will react, chemists usually simulate electrons and atoms in simplified models, a process which is computationally expensive and often inaccurate.

Now, researchers from the University of Cambridge have developed a data-driven approach, inspired by genomics, where automated experiments are combined with machine learning to understand chemical reactivity, greatly speeding up the process. They’ve called their approach, which was validated on a dataset of more than 39,000 pharmaceutically relevant reactions, the chemical ‘reactome’.

Their results, reported in the journal Nature Chemistry, are the product of a collaboration between Cambridge and Pfizer.

“The reactome could change the way we think about organic chemistry,” said Dr Emma King-Smith from Cambridge’s Cavendish Laboratory, the paper’s first author. “A deeper understanding of the chemistry could enable us to make pharmaceuticals and so many other useful products much faster. But more fundamentally, the understanding we hope to generate will be beneficial to anyone who works with molecules.”

The reactome approach picks out relevant correlations between reactants, reagents, and performance of the reaction from the data, and points out gaps in the data itself. The data is generated from very fast, or high throughput, automated experiments.

“High throughput chemistry has been a game-changer, but we believed there was a way to uncover a deeper understanding of chemical reactions than what can be observed from the initial results of a high throughput experiment,” said King-Smith.

“Our approach uncovers the hidden relationships between reaction components and outcomes,” said Dr Alpha Lee, who led the research. “The dataset we trained the model on is massive — it will help bring the process of chemical discovery from trial-and-error to the age of big data.”

In a related paper, published in Nature Communications, the team developed a machine learning approach that enables chemists to introduce precise transformations to pre-specified regions of a molecule, enabling faster drug design.

The approach allows chemists to tweak complex molecules — like a last-minute design change — without having to make them from scratch. Making a molecule in the lab is typically a multi-step process, like building a house. If chemists want to vary the core of a molecule, the conventional way is to rebuild the molecule, like knocking the house down and rebuilding from scratch. However, core variations are important to medicine design.

A class of reactions, known as late-stage functionalisation reactions, attempts to directly introduce chemical transformations to the core, avoiding the need to start from scratch. However, it is challenging to make late-stage functionalisation selective and controlled — there are typically many regions of the molecules that can react, and it is difficult to predict the outcome.

“Late-stage functionalisations can yield unpredictable results and current methods of modelling, including our own expert intuition, isn’t perfect,” said King-Smith. “A more predictive model would give us the opportunity for better screening.”

The researchers developed a machine learning model that predicts where a molecule would react, and how the site of reaction vary as a function of different reaction conditions. This enables chemists to find ways to precisely tweak the core of a molecule.

“We pretrained the model on a large body of spectroscopic data — effectively teaching the model general chemistry — before fine-tuning it to predict these intricate transformations,” said King-Smith. This approach allowed the team to overcome the limitation of low data: there are relatively few late-stage functionalisation reactions reported in the scientific literature. The team experimentally validated the model on a diverse set of drug-like molecules and was able to accurately predict the sites of reactivity under different conditions.

“The application of machine learning to chemistry is often throttled by the problem that the amount of data is small compared to the vastness of chemical space,” said Lee. “Our approach — designing models that learn from large datasets that are similar but not the same as the problem we are trying to solve — resolve this fundamental low-data challenge and could unlock advances beyond late stage functionalisation.”

The research was supported in part by Pfizer and the Royal Society.

Many important reactions related to climate and environmental processes take place where water molecules interface with air. For example, the evaporation of ocean water plays an important role in atmospheric chemistry and climate science. Understanding these reactions is crucial to efforts to mitigate the human effect on our planet.

The distribution of ions at the interface of air and water can affect atmospheric processes. However, a precise understanding of the microscopic reactions at these important interfaces has so far been intensely debated.

In a paper published today in the journal Nature Chemistry, researchers from the University of Cambridge and the Max Planck Institute for Polymer Research in Germany show that ions and water molecules at the surface of most salt-water solutions, known as electrolyte solutions, are organised in a completely different way than traditionally understood. This could lead to better atmospheric chemistry models and other applications.

Technique

The researchers set out to study how water molecules are affected by the distribution of ions at the exact point where air and water meet. Traditionally, this has been done with a technique called vibrational sum-frequency generation (VSFG). With this laser radiation technique, it is possible to measure molecular vibrations directly at these key interfaces. However, although the strength of the signals can be measured, the technique does not measure whether the signals are positive or negative, which has made it difficult to interpret findings in the past. Additionally, using experimental data alone can give ambiguous results.

The team overcame these challenges by utilising a more sophisticated form of VSFG, called heterodyne-detected (HD)-VSFG, to study different electrolyte solutions. They then developed advanced computer models to simulate the interfaces in different scenarios.

The combined results showed that both positively charged ions, called cations, and negatively charged ions, called anions, are depleted from the water/air interface. The cations and anions of simple electrolytes orient water molecules in both up- and down-orientation. This is a reversal of textbook models, which teach that ions form an electrical double layer and orient water molecules in only one direction.

Co-first author Dr Yair Litman, from the Yusuf Hamied Department of Chemistry, said: “Our work demonstrates that the surface of simple electrolyte solutions has a different ion distribution than previously thought and that the ion-enriched subsurface determines how the interface is organised: at the very top there are a few layers of pure water, then an ion-rich layer, then finally the bulk salt solution.”

Co-first author Dr Kuo-Yang Chiang of the Max Planck Institute said: “This paper shows that combining high-level HD-VSFG with simulations is an invaluable tool that will contribute to the molecular-level understanding of liquid interfaces.”

Professor Mischa Bonn, who heads the Molecular Spectroscopy department of the Max Planck Institute, added: “These types of interfaces occur everywhere on the planet, so studying them not only helps our fundamental understanding but can also lead to better devices and technologies. We are applying these same methods to study solid/liquid interfaces, which could have potential applications in batteries and energy storage.”

But that could change rapidly, the University of Michigan survey suggests.

In all, 7.5% of people between the ages of 50 and 80 have used at least one direct-to-consumer health care service from an online-only provider, according to the new findings from the National Poll on Healthy Aging (https://www.healthyagingpoll.org/).

Of those who did use such a service, most said they were driven by convenience. More than 60% of them received a prescription, mostly for a one-time treatment. But only one-third of them told their regular health care provider about the prescription.

People in their pre-Medicare years of 50 to 64 were more than twice as likely as adults over 65 to have used direct-to-consumer, or DTC, online health services (10% vs. 4%). Meanwhile, 47% of those over 65 said they had never heard of such companies.

Looking to the future, nearly a third of all older adults, and more than 42% of those age 50 to 64, said they’d be interested in using such services in the future.

The poll is based at the U-M Institute for Healthcare Policy and Innovation, and supported by AARP and Michigan Medicine, U-M’s academic medical center.

For the DTC survey, the poll team worked with members of U-M’s Center for Value-Based Insurance Design, who are interested in how cost and convenience influence people’s health care decisions and the continuity of care delivery.

A rapidly growing sector spurs concerns

The rise of DTC sites and subscription-based apps that promise convenient online access to providers who can evaluate symptoms, make diagnoses and prescribe medicines has received a lot of attention, especially amid a national crunch in availability of primary care providers and timely appointments.

Such companies include Amazon Clinic, Sesame, Roman, BetterHelp, Rosy, Lemonaid, Hims & Hers, and don’t require a referral or health insurance. Drug companies and membership-based organizations including Weight Watchers and Costco have also started offering access to such direct services.

But the trend has raised concern because of the potential for patients to receive care and prescriptions from providers who don’t know their full health history, don’t have access to their full medical records, and may not check for potentially dangerous interactions between medications.

One-third of those who had used a DTC service said their primary care provider wasn’t aware they had done so. If they received a new prescription through an encounter with a DTC health service, one-third said their regular primary care provider was not made aware of the new medication they were prescribed. The majority of those who received prescriptions through a DTC service said it was for a one-time treatment.

“These compelling findings have important implications for patient safety and continuity of care,” said Mark Fendrick, M.D., director of VBID and IHPI member who is a primary care physician at Michigan Medicine. “With rapid growth in this sector of health care predicted for this year and beyond, all providers, insurers and regulators need to pay more attention to how patients are using these services and why, and the impact on care quality and safety.” Fendrick is a professor of internal medicine in the Division of General Medicine at the U-M Medical School.

His colleague Nicole Hadeed, M.D., who also worked on the poll and is a clinical assistant professor, notes that while the number of poll participants who said they had used DTC services was relatively small, the analysis gives clues that should inform further research.

Types of care received

Nearly half of those who had used a DTC service said it has been for general health care such as treatment of allergies, sinus infections, pink eye or acid reflux, though again there was a clear divide between the 50-64 and 65-80 age groups.

Overall, nearly 12% said they’d used a service for mental health reasons, but the proportion was much higher (50%) among respondents who said they considered their mental health to be fair or poor and had used a DTC service of any kind.

As for other types of care, 15% had sought help from a DTC service for a sexual health issue, 9% had used it for skin care, 6% had used it for weight management, nearly 5% had used it for hair loss and a similar percentage had used it for pain management.

Convenience topped the list of reasons for choosing a DTC service, with 55% saying this drove their decision. But lack of access to their regular health care provider, not having a regular health care provider, or needing a service when their health provider was not open or available were each cited by around 20%. Discomfort discussing a sensitive health topic with a provider — often cited in marketing by such companies — was only mentioned by 10% of those who had turned to a DTC service.

“For both patients and providers, these findings drive home the importance of open dialogue and transparency about the potential uses, benefits and risks of these services — and the importance of maintaining contact for ongoing primary care,” said Jeffrey Kullgren, M.D., M.P.H., M.S., director of the poll and a primary care provider at the VA Ann Arbor Healthcare System who is also an associate professor at the Medical School.

More than 55% of the poll respondents who had used a direct-to-consumer service said the overall quality of care they get from their primary care provider is better than what they received from a DTC provider.

Fendrick and Hadeed wrote about the potential long-term change to primary care use from telehealth services in a piece published early in the COVID-19 pandemic in the American Journal of Managed Care.

And in fact, 58% of poll respondents who had used a DTC service had started doing so in 2020, 2021 or 2022.

The rapid pivot during the pandemic to vaccination in pharmacies, and not just primary care clinics, has also changed how people think about alternate ways of getting care that might be closer to home or have more flexible hours.

However, Fendrick notes, pharmacies share information about vaccination with insurance companies and statewide immunization registries that primary care providers can access.

“Patients will increasingly seek care online because of the convenience it can provide, especially for those willing to pay the cost out of pocket,” said Fendrick. “Its use will likely be boosted by the rapidly increasing number of online vendors and the national shortage of primary care clinicians. The recent launch of a telemedicine platform offering home delivery for the new highly popular weight loss drugs is a noteworthy example of this trend.”

He added, “Given a likely expansion of online care, it is critical that individuals inform their usual clinician and that we providers consistently ask our patients regarding their use. Similar to my routinely asking patients about which supplements, vitamins and over-the-counter medications they’re taking, it should become standard practice for me to inquire about prescriptions or diagnoses they’ve received online, as it might influence their care.”

The poll was a nationally representative survey conducted by NORC at the University of Chicago for IHPI and administered online and via phone in July and August 2023 among 2,657 adults aged 50 to 80. In all, 168 respondents reported having used a DTC health care service. The sample was subsequently weighted to reflect the U.S. population. Read past National Poll on Healthy Aging reports (https://www.healthyagingpoll.org/reports-more) and about the poll methodology (https://www.healthyagingpoll.org/survey-methods).

Key Features of Graphene Technology

Following years of research under the European Graphene Flagship project, ICN2 spearheaded in collaboration with the University of Manchester the development of EGNITE (Engineered Graphene for Neural Interfaces), a novel class of flexible, high-resolution, high-precision graphene-based implantable neurotechnology. The results published today in Nature Neurotechnology aim to contribute with innovative technologies to the blooming landscape of neuroelectronics and brain-computer interfaces.

EGNITE builds on the vast experience of its inventors in fabrication and medical translation of carbon nanomaterials. This innovative technology based on nanoporous graphene integrates fabrication processes standard in the semiconductor industry to assemble graphene microelectrodes of a mere 25 µm in diameter. The graphene microelectrodes exhibit low impedance and high charge injection, essential attributes for flexible and efficient neural interfaces.

Preclinical Validation of Functionality

Preclinical studies by various neuroscience and biomedical experts that partnered with ICN2, using different models for both the central and peripheral nervous system, demonstrated the capacity of EGNITE in recording high-fidelity neural signals with exceptional clarity and precision and, more importantly, afford highly targeted nerve modulation. The unique combination of high-fidelity signal recording and precise nerve stimulation offered by EGNITE technology represents a potentially critical advancement in neuroelectronic therapeutics.

This innovative approach addresses a critical gap in neurotechnology, which has seen little advancement in materials over the last two decades. The development of EGNITE electrodes has the capacity to place graphene at the forefront of neurotechnological materials.

International Collaboration and Scientific Leadership

The technology presented today builds on the legacy of the Graphene Flagship, a European initiative that during the last decade strived to advance European strategic leadership in technologies that rely on graphene and other 2D materials. Behind this scientific breakthrough is a collaborative effort led by ICN2 researchers Damià Viana (now at INBRAIN Neuroelectronics), Steven T. Walston (now at University of Southern California), and Eduard Masvidal-Codina, under the guidance of ICREA Jose A. Garrido, leader of the ICN2Advanced Electronic Materials and Devices Group, and ICREA Kostas Kostarelos, leader of the ICN2Nanomedicine Lab and the Faculty of Biology, Medicine & Health at the University of Manchester (UK). The research has had the participation of Xavier Navarro, Natàlia de la Oliva, Bruno Rodríguez-Meana and Jaume del Valle, from the Institute of Neurosciences and the Department of Cellular Biology, Physiology and Immunology of the Universitat Autònoma de Barcelona (UAB).

The collaboration includes the contribution from leading national and international institutions, such as the Institut de Microelectrònica de Barcelona — IMB-CNM (CSIC), the National Graphene Institute in Manchester (UK), and the Grenoble Institut des Neurosciences — Université Grenoble Alpes (France) and the University of Barcelona. The technology integration into the standard semiconductor fabrication processes has been performed at the Micro and Nanofabrication cleanroom of the IMB-CNM (CSIC), under the supervision of CIBER researcher Dr Xavi Illa.

Clinical Translation: Next Steps

The EGNITE technology described in the Nature Nanotechnology article has been patented and licensed to INBRAIN Neuroelectronics, a spin-off based in Barcelona from ICN2 and ICREA, with support from IMB-CNM (CSIC). The company, also a partner in the Graphene Flagship project, is leading the translation of the technology into clinical applications and products. Under the direction of CEO Carolina Aguilar, INBRAIN Neuroelectronics is gearing up for the first-in-human clinical trials of this innovative graphene technology.

The industrial and innovation landscape on semiconductor engineering in Catalonia, where ambitious national strategies plan to build state-of-the-art facilities to produce semiconductor technologies based on emerging materials, offer an unprecedented opportunity to accelerate the translation of such results presented today into clinical applications.

Closing Remarks

The Nature Nanotechnology article describes an innovative graphene-based neurotechnology that can be upscaled using established semiconductor fabrication processes, holding the potential for a transformative impact. ICN2 and its partners continue to advance and mature the described technology with the aim to translate it into a real efficacious and innovative therapeutic neurotechnology.

The genome or DNA of an organism incorporates the blueprint for proteins and orchestrates the production of new cells. Aiming to combat pathogens, cure genetic diseases or achieve other positive effects, molecular biological CRISPR technologies are being used to specifically alter or silence genes and inhibit protein production.

One of these molecular biological tools is CRISPRi (from “CRISPR interference”). CRISPRi blocks genes and gene expression without modifying the DNA sequence. As with the CRISPR-Cas system also known as “gene scissors,” this tool involves a ribonucleic acid (RNA), which serves as a guide RNA to direct a nuclease (Cas). In contrast to gene scissors, however, the CRISPRi nuclease only binds to the DNA without cutting it. This binding results in the corresponding gene not being transcribed and thus remaining silent.

Until now, it has been challenging to predict the performance of this method for a specific gene. Researchers from the Würzburg Helmholtz Institute for RNA-based Infection Research (HIRI) in cooperation with the University of Würzburg and the Helmholtz Artificial Intelligence Cooperation Unit (Helmholtz AI) have now developed a machine learning approach using data integration and artificial intelligence (AI) to improve such predictions in the future.

The approach

CRISPRi screens are a highly sensitive tool that can be used to investigate the effects of reduced gene expression. In their study, published today in the journal Genome Biology, the scientists used data from multiple genome-wide CRISPRi essentiality screens to train a machine learning approach. Their goal: to better predict the efficacy of the engineered guide RNAs deployed in the CRISPRi system.

“Unfortunately, genome-wide screens only provide indirect information about guide efficiency. Hence, we have applied a new machine learning method that disentangles the efficacy of the guide RNA from the impact of the silenced gene,” explains Lars Barquist. The computational biologist initiated the study and heads a bioinformatics research group at the Würzburg Helmholtz Institute, a site of the Braunschweig Helmholtz Centre for Infection Research in cooperation with the Julius-Maximilians-Universität Würzburg.

Supported by additional AI tools (“Explainable AI”), the team established comprehensible design rules for future CRISPRi experiments. The study authors validated their approach by conducting an independent screen targeting essential bacterial genes, showing that their predictions were more accurate than previous methods.

“The results have shown that our model outperforms existing methods and provides more reliable predictions of CRISPRi performance when targeting specific genes,” says Yanying Yu, PhD student in Lars Barquist’s research group and first author of the study.

The scientists were particularly surprised to find that the guide RNA itself is not the primary factor in determining CRISPRi depletion in essentiality screens. “Certain gene-specific characteristics related to gene expression appear to have a greater impact than previously assumed,” explains Yu.

The study also reveals that integrating data from multiple data sets significantly improves the predictive accuracy and enables a more reliable assessment of the efficiency of guide RNAs. “Expanding our training data by pulling together multiple experiments is essential to create better prediction models. Prior to our study, lack of data was a major limiting factor for prediction accuracy,” summarizes junior professor Barquist. The approach now published will be very helpful in planning more effective CRISPRi experiments in the future and serve both biotechnology and basic research. “Our study provides a blueprint for developing more precise tools to manipulate bacterial gene expression and ultimately help to better understand and combat pathogens,” says Barquist.

The results at a glance

• Gene features matter: The characteristics of targeted genes have a significant impact on guide RNA depletion in genome-wide screens.

• Data integration improves predictions: Combining data from multiple CRISPRi screens significantly improves the accuracy of prediction models and enables more reliable estimates of guide RNA efficiency.

• Designing better CRISPRi experiments: The study provides valuable insights for designing more effective CRISPRi experiments by predicting guide RNA efficiency, enabling precise gene-silencing strategies.

Funding

The study was supported by funds from the Bavarian State Ministry of Science and Art through the bayresq.net research network.