A Guide to Markush Structure Patent Search and Analysis

Prior art searching · Chemical structures

The prior art that keywords will never return.

In chemical patents, the most dangerous disclosure is usually one nobody named. It sits inside a generic claim — drawn once, never written out. Structure search is how you find it.

01 — Why chemical prior art is different

One drawing. Thousands of disclosures. No names.

A keyword prior art search assumes the invention was described in words. In chemistry, it usually wasn’t. It was drawn as a generic structure with placeholders, and the claim text quietly defines every variation it covers.

That’s a Markush structure: a fixed core scaffold carrying labelled positions — R1, R2, X — plus a list of what each is allowed to be. Nothing in that document holds your compound’s name, its registry number, or a drawing of it specifically. Your compound may still be squarely inside it.

What a generic claim looks like
R1
|
R3CORE SCAFFOLDR2
fixed skeleton · variable positions

…a compound of Formula I, wherein R1 is selected from CH3, CF3, C2H5, cyclopropyl; R2 is selected from Cl, F, OCH3, OH, H; R3 is selected from H, NH2, CH3.

4 × 5 × 3 = 60 compounds disclosed by one drawing.

Every one is potentially citable prior art. None is searchable by name, and only a handful will ever appear as a worked example. This is the gap structure search exists to close.

02 — The toolkit

Four structure search types, four different answers

“Structure search” isn’t one thing. It’s four — and running the wrong one is how a novelty search comes back clean on a compound disclosed a decade ago.

Search type The question it answers Blind to Recall Effort
Exact structure Has this precise compound been disclosed before? Everything adjacent — one methyl group defeats it Narrow Low
Substructure Who has disclosed anything containing our core? Analogs where the core itself was altered High Low
Markush / claim scope Does our compound fall inside somebody’s generic claim? Anything outside the claim’s defined lists, however close Scoped High
Similarity What else is chemically near this, on any scaffold? Little — the trade is noise, which scales with reach Broad Medium

Recall and effort are directional planning guidance, not measured benchmarks.

03 — See the difference

Six prior art records. Four verdicts.

One query compound, one small prior art set. Switch the search type and watch which records surface.

Search type — select one






PA-01 · US granted
Claims the query compound
Names and draws the exact molecule as a worked example. The one hit every search type agrees on.
PA-02 · WO pending
Generic claim covering it
Never names your compound — but its substituent lists cover it. Invisible to exact search; decisive for freedom to operate.
PA-03 · EP granted
Broader fused analog
Contains the query core plus an extra ring. A substructure hit, outside the claim as defined.
PA-04 · Journal paper
Same core, different group
Non-patent literature disclosing the scaffold with a substituent outside the claim lists. Still citable art.
PA-05 · CN pending
Ring-contracted analog
A likely design-around. No structural search reaches it — only similarity does.
PA-06 · US granted
Unrelated scaffold
Shares a halogen and little else. The noise a well-set threshold keeps out of your review pile.
1 of 6 returned. Clean-looking, and dangerously incomplete on its own.
4 of 6 returned. Good coverage — but it can’t tell you which of these actually claim your compound.
2 of 6 returned. The records that bear on whether you are free to operate.
5 of 6 returned. Reaches the design-around no structural method can — at the cost of a longer review list.

Note PA-02. An exact search says the compound is new. A claim-scope search finds a pending application that already covers it. That difference separates a filing decision from a clearance problem.

04 — Match method to mandate

Which search does your decision need?

The search type follows the business question, not the chemistry. Pick your mandate.

Search planner






Recommended mix

Exact and substructure, backed by claim-scope

Establish quickly whether the compound has been disclosed outright, then widen to substructure. The step teams skip is claim-scope searching — a generic claim can anticipate a compound nobody ever named.

Deliverable · novelty report with ranked art and anticipation notes

Recommended mix

Markush claim-scope, with legal status screening

Freedom to operate turns on whether a live claim covers what you intend to sell, so resemblance isn’t the test — membership is. We search generic claims directly, then filter to your markets and check status.

Deliverable · FTO report with claim-by-claim analysis and jurisdiction status

Recommended mix

Substructure and similarity against the priority date

You need art that predates their filing and reads on their claims. Broad structural coverage plus a similarity sweep finds the disclosures their examiner didn’t, including non-patent literature.

Deliverable · invalidation search with claim charts mapping art to claim elements

Recommended mix

Substructure and similarity, analysed by assignee and date

Landscaping rewards recall over precision. Cast wide across the scaffold space, then let the filing pattern show who is working where and which chemistry is still open.

Deliverable · landscape report with scaffold clustering and white-space analysis

05 — Why this takes expertise

The volume behind a single claim

Generic claims scale multiplicatively. A modest one covers more compounds than any team could review by hand — which is why claim-scope searching is indexed and tested, not read.

Compounds covered — four options per variable position
  • 1 position4
  • 2 positions16
  • 3 positions64
  • 4 positions256
  • 5 positions1,024
  • 6 positions4,096
  • 8 positions65,536
  • 10 positions1,048,576
Log scaleAmber — beyond manual review

Counts assume every listed combination is legal. Real claims are smaller — valence rules, mutually exclusive definitions and proviso language prune the total. They rarely prune it to a size you can eyeball.

06 — Scope of the search

Where chemical prior art actually sits

A structure search is only as good as the sources it runs against. Patent chemistry is spread across offices, registries and literature that index structures differently — or not at all.

Source coverage

Well indexed by structure

  • Granted patents, major offices
  • Published applications, PCT
  • Chemical registries
  • Commercial compound databases

Partial — needs expert handling

  • Markush claims in full text
  • Non-Latin-script filings
  • Older, image-only records
  • Sequence-adjacent chemistry

Text-only — structure search won’t reach it

  • Journal articles, abstracts
  • Theses, posters
  • Regulatory filings
  • Product literature

This is why a structure search alone isn’t a prior art search. Structure finds the chemistry; targeted text and classification searching finds the rest.

07 — Working with us

How a Synoptic IP structure search runs

Each stage narrows what the next has to handle, and every stage is recorded so the search is reproducible if it’s ever challenged.

  1. Brief and scope

    We agree the decision the search supports, the jurisdictions that matter, the cut-off date, and how much recall you’ll trade for a shorter review list. This conversation shapes the result more than any tool does.

  2. Query construction

    We build the query core at the smallest size that still describes the invention. Every atom added to a query is a hit you’ll never see — so this step is deliberate, documented, and iterated with your chemists.

  3. Multi-database execution

    Structure queries across patent and registry sources, run in parallel with classification and text searching for disclosure that was never indexed structurally.

  4. Expert screening

    Raw hits get read, not just counted. We check whether a generic claim genuinely reads on your compound, whether the provisos exclude it, and whether the record is live in the markets you care about.

  5. Reporting and handover

    You receive ranked, annotated art with the reasoning attached — plus the full query log, so your attorneys can reproduce or extend the search rather than start again.

What’s in the deliverable
  • Ranked artRecords ordered by how closely they read on the query, with the specific claim or passage identified.
  • Relevance notesWhy each record matters — anticipation, obviousness combination, or claim-scope overlap — in language your team can act on.
  • Structure evidenceQuery structure alongside each hit’s disclosed structure or generic claim, so relevance is visible rather than asserted.
  • Legal statusGrant status, family members and jurisdictions for anything bearing on freedom to operate.
  • Search logDatabases, query strings, structure files, filters and dates — the audit trail that makes the search defensible.

08 — Before you commission

What to send with your brief

Searches stall for the same handful of reasons. Tick these off and we can start on day one.

09 — Common questions

What clients ask us

Can’t our team just run this in a free database?

You can run exact and substructure queries in public tools, and for a quick directional read that’s often enough. What public tools don’t do well is claim-scope searching across generic claims, coverage of offices with limited structure indexing, and the expert screening that decides whether a generic claim actually reads on your compound. That’s where searches come back falsely clean.

Is a substructure hit the same as an infringement finding?

No. A hit means a record contains your fragment. Whether a claim covers your compound depends on the defined substituent lists read together with the provisos — and whether that amounts to infringement is a legal question for your counsel. We identify and evidence the art; your attorneys draw the conclusion.

Do you search non-patent literature too?

Yes, and it’s necessary. A significant share of chemical disclosure appears first in journals, theses, conference material and regulatory filings that are not structure-indexed. We pair structure queries with targeted text and classification searching to reach it.

What if we don’t have a single defined compound yet?

Common in early-stage programmes, and workable. We search the genus rather than a point compound — the core plus the substituent ranges you’re considering — which also shows which parts of your intended claim are already occupied.

Can you handle filings in Chinese, Japanese or Korean?

Yes. These are exactly the filings where structure searching earns its keep, because text searching across scripts and transliterated chemical names is unreliable. Structures are language-independent; we search them as structures and confirm relevance from the original document.

How do you keep our chemistry confidential?

Query structures for unfiled inventions are handled under NDA and searched in environments appropriate to their sensitivity. Tell us at briefing if the structure is unpublished and we’ll confirm arrangements before anything is run.

Bring us the structure. We’ll bring back the art.

Novelty, freedom to operate, invalidation or landscape — for pharma and biotech teams whose decisions turn on chemistry that keyword searching can’t see.

Talk to us about a structure search →

Synoptic IP · Prior art and structure search
Educational content — not legal advice