A patent describes a compound series relevant to your research. You want to understand how changes in structure relate to the reported activity. But the structures appear in synthetic examples, the measurements sit in results tables, and the experimental methods are described elsewhere.
The information is there, but it is not yet organized for analysis. Building a structure–activity relationship (SAR) dataset means connecting those pieces so you can work with the compounds and their results together. Patsnap Eureka’s SAR Data Extraction helps assemble them as structured data you can review and export.
Key takeaways
- Extract the links between compounds, measurements and experiments, alongside the values themselves.
- Let the research question determine which observations you take forward.
- Keep source references with the data so you can interpret results and revisit unresolved details.
What makes a patent useful for SAR analysis?
SAR work examines how changes to a molecule relate to changes in biological activity. A change in a functional group or stereochemistry may affect potency or selectivity, but interpreting that relationship requires knowing which structure was tested and how its activity was measured.
An activity table alone may not answer those questions. The compound code must connect to a defined structure, and the measurement needs its indicator, value, unit and experimental context. The structure and result may have different source locations. Bringing them together gives you an observation you can interpret, rather than a number you still need to reconstruct from the patent.
Start with the patent and your research question
Choose a patent containing the series you want to examine. Decide which reported results matter to the work: activity against a particular enzyme, a cellular response, or another property relevant to the program. This gives you a basis for selecting observations from the extraction without treating every endpoint as interchangeable.
Keep the exact patent document you intend to analyze, including the structures, result tables and experimental descriptions relevant to your question.
Bring the patent fragments into a connected record
A useful SAR record connects the compound to its measurement and experimental context. The table below shows how information from different parts of a patent fits together.
Illustrative example: one compound–activity observation
| Where the information appears | What to carry into the record | What it establishes |
|---|---|---|
| A synthetic example or structure figure | Compound code, chemical structure and structure source | Which molecule the observation concerns |
| An activity table | The corresponding compound code, indicator, qualifier, value, unit and value source | What the patent reports for that molecule |
| An experimental description | Target, experimental subject, method and relevant conditions where stated | How to interpret the reported result |
The result is one connected observation: a compound and structure, a reported measurement, its experimental context and the references supporting those links. A protein target and the cell line used in an experiment remain separate pieces of information. If the compound has several measurements, retain the observations individually so each keeps its own context.
Check the extracted structure against its patent source, including substituents and any specified stereochemistry, and correct confirmed mismatches. Match the code in the result table to that structure. Where a link remains ambiguous, keep the observation separate for follow-up.
Read values with their headings and notes. Preserve a > or < qualifier with its reported limit; it is a bound, not an exact measurement. If a unit is missing, check the table heading and footnotes. When converting units, retain the original result alongside the conversion and keep the measurement relation intact.
Put the extracted data to work
For a series review, group observations around the question you set at the start. An enzyme-inhibition IC50 and a cell-viability IC50 can both be useful for the same compound, but they describe different experiments. Keep them distinct when examining structural changes across the series. Checking assay descriptions can reveal obvious mismatches, but similar labels do not guarantee comparable experiments.
The structured data also lets teams reviewing patents or researching competitors examine disclosed compounds and results without repeatedly searching the full document. Source references let researchers return to the supporting passages when a finding merits closer investigation.
For computational work, keep experimental context and unresolved records available when deciding what belongs in a modeling dataset.
How Patsnap Eureka supports SAR extraction
Patsnap Eureka’s SAR Data Extraction supports the work of turning scattered patent information into structured records you can review and use in research:
- Submit the patent. The workflow focuses on supported small-molecule patent documents. It checks document and compound support, identifies quantitative results and checks required fields before returning structured data. A qualitative statement that a compound is “active” does not supply a numerical activity measurement.
- Work from connected data. Outputs can include compound codes, structures, activity indicators such as IC50 or EC50, values and units, standardized targets, experimental subjects, methods and dosing information. Separate structure and value sources let you trace the records to the relevant patent passages.
- Review and refine the results. Check extracted structure matches and make edits where needed. Use the source references to carry out the measurement and assay checks above, keeping unresolved observations separate from the data you take into analysis.
- Export for the next research task. Download CSV or Excel files for series review, patent research or computational analysis. Exported files can include SMILES even when those strings are not displayed in the interface; check them against the patent structures before using them downstream.
Open SAR Data Extraction with the patent and compound series you want to study.
Open SAR Data Extraction
