Design, make, test, and analyze (DMTA) is arguably the most effective discovery method, but it takes significant time and effort. While effective science requires strict adherence to the DMTA cycle, market pressures require that researchers speed the process—going from concept to product in months and years instead of years and decades.
Each team is performing a different aspect of discovery:
- Biologists are designing and running assays.
- Synthetic chemists plan and execute synthetic routes to produce compounds.
- Computational chemists are designing compounds and predicting their properties with molecular modeling and machine learning.
- Data scientists build the tools for computational chemists.
As such, the results they generate are not necessarily comprehensive.
They are all using different tools (some with AI, some not) and all are generating data across separate individual data silos. Inventory management is in one silo, experimental results are in another, and a third silo houses computational predictions. With the scattered information, no single source of truth exists.
While automation is in place in most labs, ingesting data into the electronic lab notebook (ELN) from a spreadsheet, for example, many parts of the DMTA process are slowed by the need for manual data entry and to reformat, transfer, process, and validate information across the various data silos.
While most labs have some degree of automation, many DMTA processes are slowed by manual data entry and the step-by-step process required to move information across various data silos.
Communication among registration systems, analytical tools, and software platforms is critical to creating automated data flow. This ensures that everyone (with the correct permissions) can access and use all the information in real time, with no tunnel view.
Efficient process management within the DMTA infrastructure requires a focus on:
- Reducing overall cycle time
- Creating and simplifying access to a single source of truth
- Streamlining and automating processes
- Scaling AI adoption and integration
- Accelerating analysis of ever-increasing data volumes
Together, these priorities clear the path for faster insight, better decisions, and stronger outcomes across the DMTA cycle.
The Separate DMTA Networks
While everyone is ultimately working toward a common goal, each division is using different systems and tools, relevant to its specific domain. Handing off and sharing data is usually still a manual process.
Starting with Design
The predicted activity, synthetic feasibility, physiochemical properties, and modeling results are usually shared via PDF or within presentations instead of as structured data.
Moving to Make
Chemists record experimental procedures and yields within ELNs, which aren’t always integrated with the inventory or registration system.
Fragmenting the Testing
Another challenge is accessing analytical data. Analytical instruments such as nuclear magnetic resonance spectroscopes (NMR), high-performance liquid chromatographs (HPLC), or gas chromatographs-mass spectrometers (GC-MS), generate data in vendor-specific formats, typically stored locally or within unconnected chromatography data systems (CDS).
Analyzing Parts to Create a Whole
The analytical results, synthetic details, and design predictions need to be unified to create a single coherent picture. However, all the information is stored across an unnamed number of silos. Scientists need to group all the data to gain insights on the next iteration, which could take hours, days, or months, based on the complexity of the lab research environment.
The Non-Existent Feedback Cycle
DMTA is based on feedback loops: data from test and analyze is used to fine-tune design and make/synthesize and so on. With fragmented data, the single feedback loop becomes a giant knot.
The phenomenon is particularly evident in the analytical section. Instrumental raw data (for example, spectra and chromatograms) are trapped in proprietary formats, with method parameters in the ELN. Sample-related metadata is stuck within the laboratory information management system (LIMS) and compound registration details are in the registration system.
When compounds are identified as potentially promising, reconstructing the chain to get to reproducibility is far from easy. Finding all the relevant data and correlating analytical results with molecular features to generate the next DMTA iteration requires the skills of a detective with a bit of Marie Curie and Rosalind Franklin thrown in.
Still Doing it by Hand
Instrument control and sample handling are typically the first activities automated, whereas data workflows remain manual. Each DMTA cycle transition, such as the transition from M to T, requires exporting-importing-exporting, validating, and reformatting data, which introduces potential errors and wastes time. All results from each run must be reviewed, exported, and manually uploaded to the ELN and/or LIMS, which, in high-throughput campaigns, can involve uploading hundreds of spectra and chromatograms. A misplaced decimal or a purity percentage error can lead to erroneous conclusions about compound quality.
Making DMTA AI-Ready
AI holds extraordinary promise to accelerate the DMTA cycle by automating data interpretation, optimizing reaction conditions, suggesting novel molecular scaffolds, and so on. The real value emerges when AI operates across the entire DMTA chain, learning from design outcomes, predicting synthetic feasibility, and guiding analytical priorities in a continuous feedback loop. However, if the data isn’t clean, available, and high-quality, the AI analysis can’t deliver a clear result.
Truly Enabling DMTA
What’s the real result? Version control, compliance, and traceability are major issues facing laboratories without a single source of truth. The same molecule may be in multiple systems, with different structural representations or identifiers.
Data fragmentation has significant regulatory implications for analytical chemistry groups, especially when it comes to meeting attributable, legible, contemporaneous, original, accurate (ALCOA+) standards. When analytical results can’t be traced via instrument audit trails, calibration records, and method validation data, the researchers must perform extensive reconciliation and documentation.
With a single source of truth (a single, integrated data repository), the links between compound registration, batch synthesis, and analytical characterization are easy to find and track, leading to more reliable decision-making.
For example, if a stability issue or purity deviation arises, having a single source of truth makes it easier to trace back through the reaction conditions, instrument parameters, analyst annotations, and even the design rationale.
Preparing for Integration
Simplification is key, and integration drives that. When everything can communicate automatically, workflow is simplified. No one has to upload, download, or spend time reformatting. Once the data work happens behind the scenes, workflows can be redesigned to accelerate the DMTA process itself, rather than relying on paths that move data from one place to another.
To achieve the data “simplification,” some administrative work is necessary, though. Data schemas, metadata fields, file-naming conventions, and other compound identifiers need to be standardized across the entire lab. This ensures that the data being compiled into the single source of truth makes sense to everyone. Results can be directly compared across instruments and sites. Furthermore, having instrument configurations and standardized method descriptors makes it easy to repeat any tests generated from the first DMTA cycle into the next one.
Automating Data to Achieve Faster DMTA Results
The next practical step to data connectivity and accelerating the DMTA cycle is leveraging open APIs and standardized data models. When everything is connected, benefits increase exponentially.
For example, when a compound is registered, its related structures and identifiers are accessible across the LIMS, ELN, and inventory system. During instrument analysis, raw and processed data is automatically sent to the centralized repositories, accompanied by its metadata, including operator, sample ID, and method. Meanwhile, results are checked against predefined criteria, with automated validation providing alerts if retention times deviate from norms or purity falls below the speculated rate.
Dashboards capturing the data across the entire pipeline can show real-time quality metrics for both individual projects and the entire lab itself, giving project leaders visibility into compound quality and status.
Building For the Future of Your Data
Creating a single source of truth is the first step on the path to data ascendancy. Simply combining the data you have in a single place is going to create challenges. High-throughput compound screening creates terabytes of data, and a single HPLC run produces megabytes.
That’s why structured metadata, automated data capture, and robust searchability are critical. Everything needs to be found by method, batch, compound, or researcher within seconds, not hours or even minutes. The data platform needs to manage not only experimental data but also trend analysis, process-capability monitoring, and even predictive maintenance.
DMTA, AI, and Scalability
As AI is increasingly integrated into the DMTA cycle, scalability is a challenge. The scientist-generated data across every instrument, store, and report will be systematically captured. Simultaneously, AI itself is going to generate thousands of new potential compounds.
AI can also be used to identify correlations among data, for example, chromatographic peak asymmetry and molecular polarity between “real” and “AI-generated” compounds.
When AI and the data within a single source of truth are combined, DMTA can reach never-before-possible speeds. Cycle times will decrease significantly and transcription errors will all but disappear. Full contextual visibility will lead to better decision-making. Audit trails will track everything, smoothing regulatory compliance.
One Step at a Time
Achieving faster DMTA does take time and effort, but here are simple steps to get started:
- Build champions and user engagement
- Focus on change management. Get everyone involved with the process.
- Build a clear communications and training plan.
- Map the landscape and pain points as they relate to data flows, workflows, and priorities.
- Standardize
- Create consistent file-naming conventions, metadata schemas, and compound identifiers.
- See what can be retroactively standardized so data isn’t lost.
- Go for quick wins: Integrate what’s quickest and easiest first.
Small, focused moves build momentum and set the foundation for a faster, more connected DMTA future.
Looking Forward
DMTA isn’t a dinosaur—it can be nimble and agile, mainly by managing and leveraging analytical data. The goal is to create a true closed-loop discovery system, in which design hypotheses, synthesis execution, analytical verification, and data-driven analysis lead to advances in basic science, pharma, and manufacturing across every facet where analytical chemistry meets the future.
