Use case

NMR dereplication

Use Spectra to ask whether an extract or mixture component is already known before you invest in isolation.

Target user

Natural-product and mixture chemists who see a recurring spin system and need to know if it is a known library member.

Specific problem

Dereplication is a library problem with chemistry attached. Exact-match search fails on analogues, different solvents, and incomplete 13C. Isolation campaigns then repeat last year’s compound.

Existing workflow

Compare 1H fingerprints to in-house isolates and commercial libraries. When the match is partial, the team isolates anyway.

Where time is lost

Chromatography on known flavonols, sugars or fatty-acid remnants. The cost is instrument time and lost novelty, not ranking theory.

How Spectra assists

The live product language includes dereplication explicitly: retrieve knowns and analogues, validate shifts, and leave “already known” versus “still unknown” as a human call.

Data typically required

1H at minimum; 13C when the component is concentrated enough; solvent (often DMSO-d6 or methanol); taxonomic or fraction context if you are allowed to share it.

Resulting output

A shortlist of knowns and close analogues, with residuals that show whether the extract component is “close enough to stop” or still a new analogue.

Limitations

A new analogue of a common scaffold can look like the parent until 13C or 2D data exist. Dereplication is not a novelty certificate.

Overlay with an authentic standard when available. If no standard exists, isolate only the components whose residuals stay unexplained.

Related: natural products, sample analysis.