Advancing Drug Discovery With Ai, Generative Chemistry, And 3d Molecular Design

Sponsored Content by Optibrium Ltd.Reviewed by Maria OsipovaSep 18 2025

 Discover really Optibrium is transforming early-stage supplier find done AI-powered software, generative chemistry, and 3D modelling. In this interview, Matt Segall, CEO astatine Optibrium, shares insights into nan company’s technological innovations, user-focused design, and nan early of computational supplier discovery.

Please present yourself and show america astir your inheritance and existent role.

I’m nan co-founder and CEO of Optibrium. My original inheritance is successful theoretical physics and machine science, but I moved into life sciences applications during my PhD astatine Cambridge University, wherever I studied mechanisms for supplier metabolism utilizing quantum mechanical simulations, backmost successful nan 1990s.

And past successful 2001, I moved into biotech to lead a group processing predictive modelling and decision-analysis methods for supplier discovery. This yet led to Ed Champness and maine cofounding Optibrium successful 2009. We’ve been delighted to activity pinch our teams ever since, supporting an ever-growing customer guidelines of pharma, biotech and different life sciences organizations successful their supplier and compound find objectives. Although my domiciled focuses connected guidance and strategy, I still bask moving pinch our investigation squad connected nan latest technological innovations!

Can you picture Optibrium’s ngo and really your package solutions support nan supplier find process from commencement to finish?

We create cutting-edge package and AI solutions for mini molecule design, optimization and information analysis. Our ngo is to build a broad level powered by caller technologies that amended decision-making and reside 2 of supplier discovery’s biggest challenges: ratio and productivity. 

Our platforms support nan analyzable hit-to-lead and lead optimization processes by enabling teams to extract much insights from their information and use these to creation compounds pinch a higher chance of success. This facilitates better, informed decision-making that makes nan astir of disposable time, money, and resources.

How are researchers utilizing Optibrium’s platforms to amended decision-making successful nan early stages of compound creation and optimization?

Our StarDrop™ software enables scientists to target high-quality compounds for their project’s therapeutic objectives early successful nan supplier find process, enabling faster advancement from deed to candidate. Furthermore, knowing nan structure-activity relationships successful their chemistry informs nan creation of caller compounds, backed by a broad scope of successful silico modelling and generative chemistry capabilities. This enables chemists to trial a wide scope of optimization strategies, past ore connected those astir apt to output a beardown lead aliases campaigner drug, improving nan return connected experimental investment. 

Cerella™ enables supplier find organizations to much efficaciously usage early-stage information to accurately foretell late-stage, costly outcomes, specified arsenic successful vivo PK aliases phenotypic activity. This identifies nan astir promising compounds and highlights nan measurements that will supply nan astir worth to confidently advancement compounds to experimental verification. Again, this maximizes nan return connected experimental investments.

Image Credit: A9 STUDIO /Shutterstock.com

What are immoderate of nan cardinal technological aliases technological innovations that underpin your AI-driven supplier find tools?

To target high-quality compounds, we request to harvester experimental and calculated information to measure a compound’s chance of occurrence against aggregate factors, including potency, ADME and physicochemical properties, and safety. We person pioneered nan section of multi-parameter optimization, publishing and patenting respective uniquely effective approaches.

Predicting truthful galore factors calls for a wide operation of modelling approaches. We push nan boundaries crossed methods, ranging from mechanistic models based connected quantum mechanics, which we usage to foretell supplier metabolism, done 3D molecular modelling to optimize target binding, to machine-learning methods that usage existing information to foretell caller compounds.

Beyond science, it’s really important that we present these platforms conveniently for our customers astatine a debased costs of ownership. We were delighted to partner pinch Amazon Web Services arsenic 1 of nan first technological applications for their AppStream platform. This level enables our highly ocular and interactive StarDrop exertion to beryllium accessed via a browser arsenic a afloat hosted SaaS solution arsenic good arsenic a accepted on-premises installation.

How does generative chemistry fresh into Optibrium’s broader imagination for molecular design, and what effect is it having connected real-world research?

Generative chemistry has monolithic imaginable successful supplier discovery. The sheer magnitude of chemic abstraction intends scientists simply can't research it meaningfully alone.

By combining AI and instrumentality learning pinch generative chemistry, we tin probe chemic abstraction faster and much comprehensively. This intends we tin guarantee nan champion supplier candidates are found, and valuable opportunities are not overlooked. 

But we do request much than conscionable generative chemistry to make these findings; an expert's technological knowledge and strategical knowing of nan task are basal inputs. We telephone this operation of quality expertise supported by AI algorithms successful supplier find Augmented Chemistry®. The Inspyra™ module successful our StarDrop level has a seamless feedback loop betwixt generative chemistry algorithms and an master chemist, to guideline nan algorithms to research nan astir applicable chemic abstraction to quickly place optimal compounds.

Optibrium’s devices are known for balancing rigorous subject pinch usability—how do you attack designing package that useful seamlessly for some computational chemists and broader R&D teams?

One of our specialities astatine Optibrium is making blase computational methods accessible to experimental scientists done a highly intuitive and ocular personification interface. Results must beryllium presented intelligibly and successful a measurement that’s easy to interpret, to alteration users to make effective decisions quickly. To execute this, we deliberation cautiously astir human-computer relationship and activity intimately pinch our users to understand really they deliberation astir their compounds and information successful nan discourse of their task objectives.

It's besides important to facilitate adjacent collaboration betwixt computational experts and nan chemistry and biology teams they support, and we execute this successful 2 ways. The caller collaboration capabilities coming soon successful StarDrop alteration task squad members to easy stock nan results of their analyses, helping nan full squad to use from their insights. We besides connection galore elastic ways to customize and merge StarDrop pinch in-house platforms, enabling computational chemists to incorporated their ain models and algorithms, thereby expanding their effect connected supplier find projects by making them easy accessible.

Can you locomotion america done really Optibrium’s 3D molecular modelling capabilities are helping scientists visualize and optimize compounds much effectively?

Understanding a molecule's three-dimensional building and interactions is basal to knowing its binding interactions and drug-like properties. When you only deliberation successful 2D, you tin miss this captious accusation that guides nan creation of amended campaigner drugs.

Our BioPharmics level provides industry-leading 3D ligand and structure-based creation technologies. We acquired BioPharmics successful 2023, and our colleagues and co-founders of BioPharmics, Ajay Jain and Ann Cleves, person continued to widen this exertion and show its superior capacity done published investigation and manufacture collaborations.

Benchmarks conducted by world pharma companies show that BioPharmics is much effective astatine identifying progressive compounds during virtual screening, meaning that you person to trial less compounds experimentally.

Recent investigation has achieved singular accuracy successful predicting really compounds fresh wrong macromolecule targets. This penetration enables chemists to optimise potency and properties much efficiently, reducing nan request for extended synthesis and testing. The eventual goal, however, is to foretell compound affinity—the ‘holy grail’ of computational chemistry. 

The starring technology, free-energy perturbation (FEP), useful good successful immoderate circumstances, but is constricted successful that it requires an experimentally-determined protein-ligand structure, it tin only foretell mini changes from a reference ligand and is computationally very expensive, requiring costly GPUs. The QuanSA method successful nan BioPharmics level is balanced successful accuracy, is 1000x faster without GPUs, and tin beryllium applied to overmuch larger changes successful chemic structure, moreover to different chemic series.

We’re besides pioneering applications to ‘beyond rule-of-five’ compounds specified arsenic peptidic macrocycles. Despite their therapeutic potential, nan size and elasticity of these molecules airs an tremendous situation that different molecular modelling package can’t tackle. By much quickly and rigorously exploring nan immense number of imaginable conformations of macrocycles and combining this pinch experimental biophysical data, our exertion makes effective 3D modelling of these molecules accessible for nan first time, opening nan section to nan ratio betterment this already brings to accepted small-molecule supplier discovery.

In what ways are your platforms supporting collaboration among multidisciplinary teams successful pharmaceutical and biotech organizations?

Collaboration is basal to supplier discovery. Bringing together aggregate disciplines, including chemistry, biology, and DMPK, accelerates discovery. Plus, progressively geographically dispersed teams request to activity together efficaciously to make faster progress.

To support effective collaboration, we must guarantee that everyone connected nan squad has entree to nan latest, meticulous information and insights, to forestall immoderate wasted clip pursuing ideas aliases conclusions that colleagues person already explored.

Our caller announcement of nan upcoming type 8 of our StarDrop level will embed each its creation and study capabilities successful a real-time collaboration environment. This intends that everyone connected a task is moving pinch nan same, up-to-date compounds and data, while being capable to position accusation successful a personalized, meaningful layout to execute their ain analysis. Results and decisions are past instantly shared backmost pinch colleagues.

As AI becomes progressively salient successful supplier discovery, really do you guarantee nan robustness, transparency, and technological integrity of your models?

We person nan highest standards crossed each facet of our science. Our investigation squad extensively tests and validates each caller methods earlier rolling them retired into our software. We besides prioritize technological integrity by continuing to people successful peer-reviewed journals independently and successful collaboration pinch pharmaceutical and biotech companies.

A cardinal facet of our attack to building AI models is assessing nan uncertainty successful each prediction. This accusation tin beryllium conscionable arsenic important arsenic nan consequence itself. To usage a consequence confidently successful guiding decisions, researchers request to understand conscionable really overmuch they tin spot it. Areas of precocious uncertainty tin besides uncover caller places to research and find promising opportunities that mightiness person been overlooked.

What are immoderate of nan biggest challenges your users look successful early-stage supplier discovery, and really is Optibrium helping to flooded them?

Time and cost. Drug find remains a slow and costly process. Our package addresses this by enabling teams to extract much accusation from their data, facilitating amended and much informed decision-making. This not only accelerates nan find process by reducing nan number of compounds that request to beryllium synthesized, but besides makes it importantly much businesslike by avoiding wasted experimental efforts, ruling retired those measurements that won't adhd worth earlier resources are committed.

A lesser known situation of supplier find is opportunity cost. Given nan complexity and sound successful nan information we generate, it’s each excessively easy to incorrectly discard perchance valuable campaigner drugs. Our unsocial approaches tin item imaginable missed opportunities caused by lost, uncertain, aliases incorrect experimental information for further investigation.

We besides admit that immoderate organizations whitethorn deficiency nan basal IT infrastructure to support valuable computational platforms. So we are capable to return nan support load disconnected their hands and trim nan costs of package ownership pinch a cloud-based type of StarDrop that provides nan aforesaid molecule design, optimization, and information study capabilities arsenic our desktop version. We grip each nan infrastructure, maintenance, and updates, truthful our customers tin attraction wholly connected researching and discovering caller medicines without worrying astir IT overhead.

Looking ahead, what areas of investigation aliases package improvement are astir breathtaking to your squad astatine Optibrium?

We're ever looking for caller ways to bring worth to our customers, whether that's done caller science, caller technology, aliases improved personification experience. We're moving connected further processing and enhancing our AI and instrumentality learning methods and our 3D modelling capabilities, but we're besides moving to make these incredibly powerful methods moreover much accessible to medicinal chemists whilst continuing to aboveground insights successful clear and interpretable ways for find teams.

Next year, we will beryllium starting a caller investigation task that will use 1 of nan latest areas of instrumentality learning to amended our methods of predicting really compounds will beryllium metabolized successful nan body. It has nan imaginable to drastically trim nan computational costs of moving highly meticulous simulations, which would alteration teams to tally calculations connected galore much molecules. Once developed, these models will beryllium integrated straight into StarDrop, making them disposable to biopharma companies worldwide.

Also, arsenic I mentioned earlier, we are moving quickly towards nan merchandise of StarDrop 8 astatine nan extremity of nan year, which will toggle shape really teams tin usage our package to facilitate collaboration.

Where tin our readers study more?

You tin find each nan accusation connected nan features and applications of our package portfolio connected our website astatine www.optibrium.com, aliases interaction america astatine [email protected] if you want to speak pinch 1 of our experts. We besides regularly update our Knowledge Base connected our website pinch useful content, including blogs addressing important topics successful supplier discovery, computational chemistry and instrumentality learning, lawsuit studies, publications and posters that show nan real-world effect of our technology.

You tin besides travel america connected LinkedIn to enactment updated connected our news, merchandise launches, and convention attendance.

About Matthew Segall

Matthew Segall is CEO of Optibrium. He has an MSc successful Computation from nan University of Oxford and a PhD successful theoretical physics from nan University of Cambridge. Since 2001, Matthew has led teams processing predictive models and intuitive decision-support and visualization devices for supplier discovery. Matt has published complete 40 peer-reviewed papers and book chapters connected computational chemistry, cheminformatics and supplier discovery. In 2009 he led a guidance buyout of nan StarDrop business to recovered Optibrium, which develops caller technologies and ground-breaking AI package and services, including Cerella and Inspyra, that amended nan ratio and productivity of supplier discovery.

About Optibrium Ltd.

Optibrium provides elegant package solutions for mini molecule design, optimization and information analysis. Optibrium's portfolio of products includes:

• StarDrop™, which brings assurance to nan action and creation of precocious value campaigner compounds. StarDrop creates an intuitive, highly ocular and elastic situation to facilitate and velocity up lead recognition and optimization, quickly targeting effective campaigner compounds pinch a precocious probability of occurrence downstream.
• Cerella™, a deployable AI level that creates caller worth from supplier find data, revealing hidden insights into nan champion compounds and astir valuable experiments for each project. Cerella makes assured predictions, accurately filling successful missing values, particularly for costly downstream experiments that can’t beryllium predicted by different methods. This enables teams to do more, moreover pinch sparse, constricted information sets.
• BioPharmics™, a level for fast, accurate, and robust 3D modelling from mini molecules to ample macrocycles. With industry-leading ligand- and structure-based creation capabilities, BioPharmics quickly generates meticulous conformational ensembles, predicts bound ligand poses and binding affinities without requiring macromolecule structural information, and enables high-performance virtual screening astatine scale.

Founded successful 2009, Optibrium continues to create caller products and investigation caller technologies to amended nan ratio and productivity of nan supplier find process. Optibrium useful intimately pinch its wide scope of customers and collaborators, including starring world pharma, agrochemical and flavoring companies, biotech and world groups.