Sponsored Content by TTP plcReviewed by Maria OsipovaJul 15 2026
In this interview, Wenshu Xu discusses how to overcome purification bottlenecks in modern drug discovery with advanced purification strategies, streamlined workflows, and automation.
Artificial intelligence (AI) is accelerating many aspects of drug discovery. Why has purification become such an important topic?
Artificial intelligence is enabling drug discovery teams to design and synthesize more compounds than ever before. However, as the throughput of the design and analysis stages in the design, make, test, and analyze (DMTA) cycle continues to increase, one critical component of the cycle has not advanced at the same pace: purification.
Although chromatography itself has become highly automated, many of the surrounding purification steps still rely heavily on manual intervention. These manual processes create a bottleneck that limits both the speed and scalability of modern drug discovery workflows.
What is driving the renewed momentum in small molecule drug discovery?
Small molecule drug discovery is experiencing a resurgence, fueled by advances in AI and automation that are rapidly expanding the accessible chemical space.
Compared with biologics, small molecules continue to offer several important advantages. They are easier to administer, less expensive to manufacture, and generally provide better patient compliance. Recent FDA approvals reflect this continued importance, with small molecules maintaining a central position in therapeutic development.
How is AI changing the way new molecules are designed and synthesized?
Artificial intelligence is also transforming molecular design and synthesis. Rather than relying solely on traditional retrosynthetic approaches, data-driven methodologies are enabling the development of synthesis routes that are both more practical and more representative of real experimental conditions.
At present, approximately 106 catalog molecules are accessible worldwide. As AI continues to identify new synthetic pathways and capitalize on simpler, scalable chemistries, this number could grow to 108 or even beyond. This rapid expansion is placing unprecedented demands on downstream discovery workflows.
Despite these advances, where does the DMTA cycle still struggle?
Although AI-driven molecular design, high-throughput experimentation, and data analysis have progressed rapidly, automated synthesis (the make stage of the DMTA cycle) continues to present significant challenges.
Several aspects of the make stage remain difficult to optimize, but purification has proven particularly resistant to improvement. Despite continued advances in synthetic chemistry and the growing ability of biological assays to tolerate impurities or reaction mixtures, conventional purification often remains necessary to produce reliable and reproducible experimental results.
How is purification typically performed in medicinal chemistry compared with industrial-scale manufacturing?
At industrial manufacturing scale, purification generally relies on robust and highly scalable techniques such as distillation, crystallization, and liquid-liquid extraction.
Medicinal chemistry, however, operates on milligram- and gram-scale quantities, where purification primarily depends on two techniques: solid-phase extraction (SPE) for crude sample cleanup and flash chromatography for high-resolution separation. In both cases, the separation process itself has already reached a high degree of automation.
Solid-phase extraction can readily be incorporated into robotic workflows, while modern flash chromatography systems are capable of automatically optimizing solvent gradients and fraction collection, allowing efficient separations with minimal operator involvement.
If chromatography is already highly automated, where does the real challenge lie?
The challenge no longer lies in molecular separation itself. From a workflow perspective, purification continues to function as a largely independent operation positioned between synthesis and biological testing, rather than as part of a seamless automated process.
This limitation becomes particularly evident in modern medicinal chemistry laboratories. While automated flash chromatography systems can often perform separations with very little operator input, chemists still devote considerable time to preparing samples, drying collected fractions, and transferring materials between different instruments.
Consequently, the primary bottleneck frequently arises not from chromatography itself, but from the manual handling steps that surround it.
Why are these manual handling steps so difficult to automate and what challenges do they present?
Crude reaction mixtures must be transferred, concentrated, adsorbed onto silica, and loaded onto chromatography columns. Although these operations appear relatively straightforward, they become extremely sensitive when performed at the small scales typical of medicinal chemistry.
Standardizing these processes is challenging because they often depend on the practical experience and tacit knowledge of skilled chemists. Even seemingly minor inefficiencies, such as inconsistent drying or material losses during transfer, can significantly affect both product yield and data quality.
As a result, human intervention remains necessary even within highly automated workflows, interrupting the continuity of the overall system.
This creates a fundamental imbalance across the DMTA cycle. While synthesis can frequently be parallelized and completed within a matter of hours, purification and drying often require several days, along with specialized equipment, dedicated operator time, and expert knowledge. Increasingly, the bottleneck stems from materials handling and workflow integration rather than from the molecular separation process itself.
Every manual operation introduces additional delays, variability, and resource demands, all of which become increasingly significant as DMTA throughput continues to grow.
As organizations move toward evaluating thousands rather than hundreds of compounds, purification has the potential to become a disproportionate constraint on overall cycle time, limiting the benefits that AI-driven molecular design and automated synthesis can deliver.
The future of purification will not be shaped solely by improvements in chromatography. Instead, progress will depend on rethinking how compounds are prepared and transferred, and, in some cases, whether purification is necessary at all.
What opportunities exist to automate sample preparation before chromatography? How would this work?
One promising opportunity is the automation of dry-loading sample preparation for flash chromatography.
This process involves producing a dried, silica-bound version of the crude reaction product that can be introduced directly onto a chromatography column without requiring manual handling. Achieving this objective requires the seamless integration of three key operations: combining the crude product with silica, drying the resulting mixture, and loading the prepared material into the flash chromatography system.
There is considerable potential to draw inspiration from adjacent industries. Technologies such as spray drying, encapsulation, and powder-processing techniques have long been used in pharmaceutical manufacturing and the food industry to convert liquids into stable, easy-to-handle solid materials.
These technologies were developed for industrial-scale production and would need to be adapted for medicinal chemistry applications, where even minimal material losses can have a significant impact on experimental outcomes.
Can you provide an example of how one of these technologies might be adapted?
Spray drying provides a useful example. Within pharmaceutical manufacturing, it is routinely used to convert liquid formulations into stable powder products at industrial scale.
Although medicinal chemistry operates on a much smaller scale, the underlying concept remains highly relevant. Transforming difficult-to-handle liquid samples into standardized solid materials that can be transferred easily could significantly improve automation within purification workflows.
This naturally raises a broader question: is it even necessary to generate loose solid material?
One possible alternative would be to absorb and dry crude reaction products onto structured solid supports such as silica wool, porous sponges, or other inert matrices. These materials may be easier to manipulate and could potentially be loaded directly into chromatography columns.
Similarly, techniques such as centrifugal drying may provide an effective method for controlled solvent removal, although efficiently recovering and transferring the dried material continues to present a technical challenge.
Beyond improving sample handling, are there opportunities to fundamentally change purification workflows?
Yes. The industry is gradually moving away from traditional batch-processing approaches toward more integrated continuous workflows.
Flow chemistry combined with in-line purification technologies, including membrane-based separation methods, offers the possibility of integrating synthesis and purification into a continuous process. Rather than treating purification as a separate operation, it becomes an integral part of the reaction stream, eliminating many of the discrete handling steps found in conventional workflows.
Are there examples of similar concepts already being used in drug discovery?
Comparable principles are already being demonstrated in highly parallel discovery platforms. DNA-encoded library technologies, for example, allow millions of compounds to be synthesized and screened while preserving compound identity through DNA tags rather than relying on traditional isolation and purification techniques.
Although these technologies are not direct replacements for conventional medicinal chemistry workflows, they illustrate how future DMTA systems may depend less on discrete purification steps and instead rely on more integrated, information-rich processes.
Some of the more disruptive approaches push this concept even further. Solid-phase and encoded chemistries, and droplet-based technologies make it possible to closely integrate synthesis and biological testing, frequently eliminating the need for compound isolation.
In certain early-stage discovery applications, there may even be circumstances in which purification is not required at all.
When highly efficient one-pot reactions are combined with well-characterized chemistry, it may become feasible to proceed directly from synthesis to biological evaluation. Provided that reaction conditions and individual components are well understood, experimental outcomes can later be deconvoluted retrospectively, reducing the need for purification before biological testing.
How do you see the role of purification evolving within future DMTA workflows?
The pharmaceutical industry has largely solved the challenge of separating molecules. The next major objective is developing more effective ways to handle, transfer, and integrate those molecules within increasingly automated discovery environments.
As artificial intelligence continues to increase the number of compounds entering the DMTA cycle, inefficiencies surrounding purification have the potential to become an increasingly significant limitation on throughput and overall productivity.
Improving chromatography workflows alone will not be sufficient to address this. The broader objective is to create tighter integration between synthesis and biological testing.
As purification becomes increasingly automated, more selective, or, in certain applications, unnecessary, the traditional boundary separating the make and test stages begins to disappear.
Rather than functioning as independent activities connected by manual handoffs, future DMTA workflows are likely to operate as continuous systems in which compounds move rapidly from synthesis into biological evaluation and data generation.
What changes will organizations need to make to support this transition?
Successfully making this transition will require new ways of thinking about materials handling, workflow design, and the overall role that purification plays within drug discovery programs.
In some situations, purification will continue to be an essential part of the process. In others, it may be postponed until biological activity has been confirmed or replaced entirely by alternative approaches that allow chemistry and biology to operate in much closer coordination.
Organizations that successfully address these issues will achieve far more than simply removing a laboratory bottleneck.
They will enable higher-velocity DMTA workflows, generate substantially larger volumes of high-quality experimental data, and establish the operational foundation needed to fully realize the potential of AI-enabled drug discovery.
Rather than acting as a constraint that limits efficiency, purification will become an integrated component of a high-velocity discovery engine capable of transforming ideas into experimental data faster than ever before.
What advice would you give to organizations experiencing purification-related bottlenecks? How does TTP support organizations looking to modernize their drug discovery workflows?
As compound volumes continue to increase, manual purification workflows can rapidly become a major obstacle to productivity.
TTP collaborates with pharmaceutical and biotechnology companies to develop automated chemistry, purification, and workflow integration technologies that reduce manual intervention while accelerating the transition from compound synthesis to biological insight.
The drug discovery tools team develops bespoke technologies that enable faster experimentation, higher-quality data generation, and more efficient decision-making.
Drawing upon expertise in biology, chemistry, automation, microfluidics, software, and instrumentation, the team works with clients to solve complex challenges involving synthesis, purification, screening, organoid systems, assay development, and integrated laboratory workflows.
Whether the goal is improving an existing workflow or creating an entirely new discovery platform, TTP partners with clients to increase DMTA iteration velocity and unlock the full potential of AI-enabled drug discovery.
References and further reading
- de la Torre, B.G. and Albericio, F. (2026). The Pharmaceutical Industry in 2025: An Analysis of FDA Drug Approvals from the Perspective of Molecules. Molecules, 31(3), p.419. DOI: 10.3390/molecules31030419. https://www.mdpi.com/1420-3049/31/3/419.
- BUCHI. Mini Spray Dryer S-300 | BUCHI. Available at: https://www.buchi.com/en/products/instruments/spray-dryer-s300.
- Jones, D. and Godek, E. (2026). Development, optimization, and scale-up of process parameters: Wurster coating. Developing Solid Oral Dosage Forms, pp.1333–1356. DOI:10.1016/b978-0-443-34156-4.00025-x. https://www.sciencedirect.com/science/chapter/edited-volume/abs/pii/B978044334156400025X?via%3Dihub.
- Biotage (2026). Biotage® V-10 Touch | Fast and automated evaporation system. Available at: https://www.biotage.com/products/v-10-touch-evaporation-system.
- Thermo Fisher Scientific. (2026). SpeedVac Vacuum Concentrators | Thermo Fisher Scientific - US. Available at: https://www.thermofisher.com/uk/en/home/life-science/lab-equipment/speedvac-vacuum-concentrators.html.
- Hartman, R.L., McMullen, J.P. and Jensen, K.F. (2011). Deciding Whether To Go with the Flow: Evaluating the Merits of Flow Reactors for Synthesis. Angewandte Chemie International Edition, 50(33), pp.7502–7519. DOI:10.1002/anie.201004637. https://onlinelibrary.wiley.com/doi/10.1002/anie.201004637.
- TTP (2026). Click chemistry transforming life science applications. TTP. Available at: https://www.ttp.com/insights/only-a-click-away-click-chemistry-transforms-life-science-applications.
Acknowledgments
Produced using materials originally authored by Wenshu Xu.
About Wenshu Xu
Wenshu Xu is Head of Drug Discovery Tools at TTP, where she leads the development of innovative technologies designed to accelerate drug discovery through advanced automation, high-throughput experimentation, and autonomous laboratory systems.
Her work focuses on enabling faster, more efficient DMTA workflows by bridging artificial intelligence and experimental science. Wenshu has extensive experience in developing bespoke instrumentation, microfluidics, automation platforms, and next-generation screening technologies that support pharmaceutical and biotechnology innovation.
She is particularly interested in the concept of "physical AI," where computational predictions are rapidly validated through automated wet-lab experimentation. Her expertise spans drug discovery technologies, workflow optimization, and laboratory automation, helping organizations generate high-quality experimental data at scale.
Through her work at TTP, Wenshu is helping to shape the future of autonomous drug discovery, in which AI, automation, and advanced analytics work together to accelerate the development of new therapies.
About TTP plc
At TTP we work with start-ups through to global corporates across Defense, Energy & Industrials, MedTech, Life Sciences, Satellite and Space, solving complex challenges rooted in technology.
By combining passion and flexibility with deep expertise in science, engineering and design, we help our clients unlock opportunities that make brilliant things happen. We’re independent, free-thinking and agile to the core, and for nearly 40 years that’s helped us find better solutions, faster - time and time again. Our multidisciplinary teams are shaped around the needs of each project, enabling us to tackle problems holistically from the outset. We work alongside you as one team, committed to your success as much as you are.
TTP's new spin-out Cellular Highways to commercialize advanced cell-sorting technology