US 11402387

I'll run searches to check for litigation/docket activity and current USPTO status, then synthesize with the authoritative full text provided.

Let me try to retrieve the actual claim set, since the provided document text does not include a claims section.

Search summary and confidence statement

I searched for this patent number against public web/docket-accessible sources and reviewed the authoritative full text supplied for US11402387. Key caveats up front:

• I could not locate any litigation or CAFC 2026 docket activity for US 11,402,387. Searches for 11402387 plus litigation/CAFC terms returned only unrelated matters (e.g., Eight Sleep Inc. v. Orion Longevity Inc., C.D. Cal. 2:26-cv-02460, asserting U.S. 12,377,240 and 12,370,339; a Mullen Industries v. Apple N.D. Cal. case; various PTAB IPRs on unrelated patents, e.g. IPR2025-00145 on U.S. 11,865,287 and IPR2024-01261 on U.S. 11,406,402). None involved US 11,402,387 or Brandeis. I cannot affirmatively state the patent is litigation-free; I can only say no 2026 CAFC or district-court activity for this number appeared in my searches.
• The document text I was given is the Google Patents full-text rendering (description, abstract, figures) — it does not include the literal claim set. My claim overview below is therefore derived from the SUMMARY section (which mirrors the independent claims) plus the abstract, and should be treated as a plain-language paraphrase, not verbatim claim language. I was unable to pull the exact claim text from a primary source in this session.

Bibliographic data for US 11,402,387 B2

Note on the PDF OCR showing "BRANDELS UNIVERSITY" — the Google Patents record and the assignment record both render the assignee as Brandeis University; the printed original appears to be an OCR artifact. I am flagging it rather than silently correcting the patent text.

Abstract (as given)

The method includes acquiring a mass spectrum of a molecule with peaks corresponding to a precursor ion and fragment ions. At least a portion of the fragment ions are identified as corresponding to one or more monomer subunit ion of the precursor ion by appending one or more fragment ions to an "inferable constituent" to produce a topology building block. That building block is stored in a candidate pool if the combined mass of the inferable constituent plus the fragment ion(s) satisfies a first user-defined mass tolerance. One or more candidate topologies of the precursor ion are then obtained by combining a plurality of the topology building blocks that satisfy a second user-defined mass tolerance for the precursor ion.

Plain-language overview of the independent claims (paraphrase of the SUMMARY aspects)

The SUMMARY presents four independent-claim categories:

1. Method of determining molecular structure using a mass spectrometer (method, broadest framing).

• Acquire a mass spectrum of a macromolecule containing a precursor ion (having a first m/z) and fragment ions.
• Identify one or more fragment ions to produce a candidate set comprising monomer subunit ions that are combinable with one or more "inferable constituent."
• The candidate set must include at least one candidate whose mass matches the precursor's first m/z within a selected mass tolerance.
• Reconstruct a topology for the precursor ion that falls within the mass tolerance.

2. Method for determining a topology of a molecule using a mass spectrometer (method, more specific; adds ranking).

• Acquire the mass spectrum (precursor + fragment peaks).
• Identify at least a portion of the fragment ions as corresponding to one or more monomer subunit ion of the precursor, where identification is done by (a) appending fragment ion(s) to an inferable constituent to create a candidate topology building block, and (b) storing that building block in a candidate pool if the combined mass of the inferable constituent and the fragment ion(s) satisfies a first user-defined mass tolerance.
• Produce one or more candidate topologies by combining multiple topology building blocks satisfying a second user-defined mass tolerance for the precursor ion.
• Rank the candidate topologies by a candidate topology score and select the highest-scoring topology. (Disclosure ties this scoring to a machine-learning classifier, "IonClassifier," over a contextual-feature window — e.g., 105 Da.)

3. Mass spectrometry unit (apparatus claim).

• An inlet port to receive a sample containing a macromolecule having monomer subunits.
• An ion source to ionize the sample into a precursor ion (first m/z).
• A mass analyzer to dissociate a portion of the precursor ions into fragment ions and to separate a fraction of precursor and fragment ions.
• A detector producing detection signals for that fraction.
• A controller receiving the detection signals and programmed to: (a) acquire the mass spectrum (precursor + fragment peaks); (b) identify fragment ions as monomer subunit ion(s) by the append-to-inferable-constituent / candidate-pool / first mass-tolerance procedure; and (c) reconstruct candidate topology(ies) by combining building blocks within the second mass tolerance.
• The specification describes a triple-quadrupole-style architecture (first quadrupole mass filter 410, collision cell 412 with multipole ion guide 414 and gas supply 416, second quadrupole mass filter 418), power units 430/432/434, and controller 422 with display, input devices, and data processor. It also notes alternatives such as FT-ICR, and generally any MS capable of ionizing and separating by m/z (AMS, GC-MS, LC-MS, ICP-MS, IRMS, MALDI-TOF, SELDI-TOF, tandem MS, TIMS, SSMS).

4. Method for determining a topology using a computer system (computer-implemented method).

• Provide an acquired mass spectrum (precursor + fragment peaks) to a computer system.
• The computer system is programmed to perform the same identifying step (append fragment ion(s) to an inferable constituent → topology building block → candidate pool if within the first user-defined mass tolerance) and the same reconstructing step (combine building blocks within the second user-defined mass tolerance).

Technical substance behind the claims (for context)

• Branded in the specification as "GlycoDeNovo." Two named algorithms: PeakInterpreter (builds an "interpretation-graph" bottom-up from the lightest peak, attaching up to four branches to a monosaccharide root, with complexity O(|G|×N^(H+1))) and CandidateSetReconstructor (recursively reconstructs only the topologies needed to explain the precursor ion, with mass-accuracy constraint τ).
• Distinctive stated advantages: delayed topology reconstruction (avoids reconstructing sub-topologies that never lead to the precursor), polynomial rather than exponential complexity versus brute-force/database-free search, and avoidance of the NP-hard Peak Assignment Problem.
• Handling of experimental realities: computationally adding artificial complementary peaks, permethylation, ¹⁸O reducing-end isotope labeling (2.004 Da shift), deutero-reduction (17.038 Da shift) for sialic-acid-containing glycans, and missing-cleavage tolerance by considering monosaccharide pairs.
• Ranking uses the IonClassifier machine-learning model (boosted decision trees) trained on experimental tandem MS of known glycan standards to distinguish B/C ions from Y, Z and "O" ions, described as improving true-topology rank (e.g., ¹⁸O-labeled Man9 N-glycan from 205th to 1st).

Specific uncertainties I want to flag

1. Independent claim count and exact language: not verified. I did not obtain the literal claim set. The four SUMMARY aspects above are a reliable indicator of the independent claims, but dependents, transitional phrases ("comprising"), and any narrowing language are unverified.
2. Litigation/CAFC status: negative search result only. No US 11,402,387 matter surfaced in my searches of 2026 dockets; absence of evidence in these search results is not evidence of absence.
3. Inventorship discrepancy worth noting: the published GlycoDeNovo paper (Hong, Sun, Sha, Pu, Khatri, Yu, Tang, Lin — Analytical Chemistry/PMC5647224) lists eight authors, but the patent record names only Pengyu Hong as inventor/assignor. I report the patent record literally; I make no representation about correct inventorship.
4. A related later family member appears to exist: a 2025 Brandeis publication titled "System and method for determining glycan topology using de novo glycan topology reconstruction techniques" (US 2025/0061975 A1), which discusses a "PeakInterpreter2" variant. I found this only in a search result snippet and have not verified it against a primary source; treat it as a lead, not a confirmed family relationship.
5. Priority-date and legal-status entries are expressly labeled by the source as assumptions, not legal conclusions (standard Google Patents caveat), and the recorded expiration of 2038-12-30 reflects patent-term adjustment.

I'll search for litigation involving US patent 11402387 across patent litigation sources.

Let me search more specific litigation databases for this patent.

Litigation Search Results for U.S. Patent No. 11,402,387

Finding: No litigation is known involving U.S. Patent No. 11,402,387.

I searched patent-litigation sources (Unified Patents litigation portal, CourtListener/Recap, Justia Dockets, Docket Alarm, PTAB/PTACTS records, and the Google Patents "Litigation" section for this patent). No district-court case, Court of Appeals for the Federal Circuit appeal, ITC investigation, or PTAB proceeding (IPR/PGR) asserting or challenging U.S. Patent No. 11,402,387 was found.

What the record does show for this patent

The patent's public record reflects prosecution and ownership activity only — no adversarial proceedings:

The only reassignment entries are the inventor's assignment to Brandeis and a confirmatory license to the National Institutes of Health — neither is litigation.

Important cautionary notes

1. Search results must be interpreted literally. My queries for "11402387" also surfaced unrelated items — e.g., a Japanese patent abstract citing "JP 11402387 A" (a 1987 Kawasaki Steel document) and various other patents. These are not U.S. Patent No. 11,402,387 and should not be conflated with it.

2. Unrelated Brandeis litigation exists but involves different patents. Brandeis University (with GFA Brands, Inc.) was a plaintiff in 2012–2013 patent suits concerning U.S. Patent No. 5,843,497 (the "Smart Balance" oil-blend technology, e.g., Brandeis University v. East Side Ovens, Inc., N.D. Ill. No. 1:12-cv-01508, consolidated with 12-cv-1509 through 12-cv-1513; originally filed in W.D. Wis.). Those cases are not related to the '387 patent, which issued nine years later.

3. What "no litigation found" does and does not mean. This reflects the sources searched as of the current date. A patent held by a university and directed to a glycomics/bioinformatics research method is far less likely to be asserted in infringement litigation than commercial patents, which is consistent with the absence of hits. It remains possible that (a) very recent filings are not yet indexed, or (b) a case exists that is not captured by the free databases I searched. If you need a litigation-grade negative, I'd recommend running a paid Docket Navigator / Lex Machina / PACER full-text search keyed precisely to patent number 11,402,387 and to the assignee "Brandeis University" as a plaintiff.

Bottom line

If you are looking for known litigation involving U.S. Patent No. 11,402,387 — there is none known. As of this search, the patent has not been asserted or challenged in any identified U.S. court or PTAB proceeding; its file reflects only ordinary prosecution, ownership transfer to Brandeis University, and an NIH confirmatory license.

Proceedings overview

There are zero (0) AIA trial proceedings on file against US 11,402,387 — no IPR, no PGR, no CBM — so the breakdown is 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution-denied, and the bottom line for a defendant is the inverse of a hardened patent: all 19 claims are untested at the PTAB, nothing is canceled, and you have no ready-made invalidity judgment to lean on — but you also inherit no § 315(e)(2) estoppel, meaning the entire prior-art universe remains available to you.

No proceedings to itemize

The structured "PTAB proceedings on file" block (USPTO Open Data Portal, most recent ingest) returned no AIA trials, and my independent web checks surfaced no petition, institution decision, FWD, or appeal naming this patent or patent owner. Because there is no proceeding, the required per-proceeding fields (petitioner, filing date, panel, grounds, institution reasoning, FWD claim-level dispositions, settlement terms, CAFC docket) have no content and I will not invent any. Reporting them would violate the no-fabrication constraint.

Negative-finding log (what I checked, and what came back empty):

One caveat worth stating plainly: this is a negative finding, not proof of absence. Public web indexing of PTAB dockets is incomplete and lagged. Before relying on "no PTAB activity," a defendant should re-run the check directly at the Board's own systems: PTAB E2E / PTAB Center and the USPTO Patent Center file wrapper for app. 16/616,831 (which would show any petition as a third-party paper). Public docket aggregators (CourtListener, Docket Alarm, RPX Insight) are useful for the parallel question of whether the patent is being asserted.

Strategic summary

Claim status: the entire claim set is UNTESTED. US 11,402,387 issued 2022-08-02 with 19 claims. Claims 1 and 14 are the independents (claim 1 a method; claim 14 a mass spectrometry unit). No claim has ever been held unpatentable, canceled by certificate, disclaimed, or confirmed by the Board. Claim 13 recites the "0.02 Da or less" mass-tolerance limitation, and claims 10–12 capture the machine-learning "contextual features" / "approximately 105 Da" mass-difference window — this is the crown-jewel claim family, and it is fully intact. For a defendant, the practical consequence is that there is no partial-validity safe harbor and no claim you can treat as already dead. Every claim you are accused of infringing must be attacked from scratch.

Estoppel landscape: you are unencumbered. Section 315(e)(2) estoppel only attaches to a petitioner (and its RPIs/privies) after an IPR results in a final written decision under § 318(a). With no FWD, there is no estoppel running against anyone — not against you, not against any would-be co-defendant, and not against any manufacturer or customer upstream of you. Conversely, the patent has never been through an adversarial validity test, so its prosecution record is the only thing shaping it. That record is where the realistic § 112 and art-based theories live. Two specific hooks from the face of the patent worth developing: (i) the priority chain runs to provisionals 62/513,495 (2017-06-01) and 62/531,229 (2017-07-11), with the PCT filed 2018-06-01 — any claim element not supported in those provisionals gets a 2018-06-01 effective date, which materially widens the § 102/§ 103 art space; and (ii) the applicants' own GlycoDeNovo publication (Hong et al., J. Am. Soc. Mass Spectrom. 2017, 28(11):2288–2301) is the closest art to the disclosed algorithm — but it is also the inventors' own work and appears to postdate or fall within the grace period of the 2017-06-01 provisional, so a § 102(b)(1)(A) exception almost certainly defeats it as a standalone ground. The stronger art attack is the secondary literature the patent itself cites (e.g., Saar-Tsechansky et al., JMLR 8:1625–1657 (2007), cited on the face; US 2011/0137570 A1 to Lapadula; US 2015/0340213 A1 to Belov; WO 2009/154964 A2; CN 106404883 A) combined with known de novo sequencers (STAT, StrOligo, GLYCH, GlycoMaster, glyfon) discussed in the specification's own background.

Pattern signals — the interesting one is portfolio, not litigation. There is no repeat-petitioner pattern (no petitioner at all), no patent owner appeal activity (there is nothing to appeal), and no defensive aggregator (Unified Patents, RPX, Askeladden, Open Invention Network) in the chain. What is observable is that the owner is still building: a continuation in this family published as US 2025/0061975 A1 ("System and method for determining glycan topology using de novo glycan topology reconstruction techniques" — the GlycoDeNovo2 work, with composition filtering and empirical p-values). That is a real defensive-planning fact. A challenge to the '387 buys you nothing against newly issuing claims in that continuation, and if you are negotiating a license you need to scope it to the family, not the single patent. Also note the funded-research trail: NIH grant numbers P41 GM104603 and U01 CA221234 (government rights — the government holds a nonexclusive license), and a 2023-09-08 NIH confirmatory license recorded against the patent. That is a licensing/march-in-rights consideration, not an invalidity one.

Why the silence is itself a signal. A patent with a broad method claim over mass-spec data processing in the glycomics space, issued in 2022 to a university, has now gone ~4 years without an IPR. That usually means one of three things: (a) it has never been asserted (the most likely explanation, given the absence of any litigation trail I could find); (b) it has been asserted only in settlement-driven campaigns where defendants priced the risk below the cost of a petition; or (c) potential petitioners concluded the art was too thin. Each implies a different problem for you. If you have received a demand letter citing the '387, you would be the first challenger — which is a mixed blessing: no estoppel and no adverse precedent, but also no roadmap of what art works, and § 315(b)'s one-year clock starts on service of a complaint alleging infringement.

Recommended next steps

1. Confirm the negative finding at the source before you act on it. Pull the file wrapper for application 16/616,831 in USPTO Patent Center and search the patent number in PTAB E2E / PTAB Center. If nothing appears, treat "no PTAB activity" as confirmed and update your risk memo accordingly. Re-check quarterly — nothing stops a petition from being filed tomorrow.
2. Do not assume claim 1 or claim 14 is weak just because it is untested. There is no FWD to quote, so there is no disposition to cite and no cancellation to leverage. Any argument that "the claims are already invalid" would be unsupported. Your invalidity theory must be built from scratch on the pre-2017-06-01 art plus the priority-date question above.
3. Calendar the statutory clocks if a complaint lands. IPR is available at any time on § 102/§ 103 grounds over patents and printed publications; the PGR window closed 2023-05-02 (nine months after the 2022-08-02 issue date), so PGR — and with it § 112 and § 101 challenges at the Board — is no longer available. That is a genuine loss: the "user-defined mass tolerance," "inferable constituent" and machine-learning claim language reads like fertile § 112 written-description ground, and § 112 is now a district-court / ITC-only theory. Once served, your § 315(b) one-year bar clock starts; trial-stage deadlines then run 6 months to institution (statutory, from petition filing plus any § 315(b) extension) and the Board's FWD is due within one year of institution under § 316(a)(11).
4. Scope any enforcement response to the whole family, not the '387. Check the status of US 2025/0061975 A1 (and any other continuations) in Patent Center. Claims issuing there will not be covered by a challenge to — or a settlement of — the '387.
5. If you want to challenge, file before you are sued if you can. With no first-filer on this patent, filing a pre-emptive IPR before service avoids the § 315(b) bar entirely and lets you pick the art and the panel timing; if you are already a defendant, expect the Board to consider Fintiv-type discretionary denial if a parallel district court trial date is set before the FWD deadline.
6. Coordinate with co-defendants and upstream suppliers on real party in interest. Because no estoppel exists yet, the first petitioner's filing decision materially affects everyone downstream (RPIs and privies get bound by the resulting estoppel). Get the RPI/privity analysis right before filing, not after.

Sources consulted (for the negative finding): US 11,402,387 B2 on Google Patents; US 11,402,387 on Justia; the structured "PTAB proceedings on file" block supplied in the prompt (USPTO ODP, 0 proceedings); targeted web searches for PTAB and Federal Circuit activity (none found).

Explicit limitations: I found no PTAB proceeding, no FWD, no settlement, and no CAFC appeal to cite — so I have cited none, rather than supplying placeholder numbers. No judge panel, filing date, ground, or disposition is stated anywhere above, because none exists in the record as of this review. Date format throughout is YYYY-MM-DD. If a proceeding does exist beyond public indexing, its details can only be confirmed through PTAB E2E / Patent Center, not from anything available to me here.

I'll research the assignment record, correspondent of record, and any NPE indicators for US 11,402,387.

Ownership & Assignment Analysis — US 11,402,387 B2

Primary-source caveat up front. I attempted to pull the assignment records directly (USPTO Patent Assignment Search / Assignment Center, keyed to patent 11,402,387 and application 16/616,831). My queries returned no reel/frame-level data, and no correspondent-of-record entries. Everything below in the Assignment timeline section is therefore built from the Google Patents legal-events table supplied in the authoritative full text (which enumerates the same recorded events but omits reel/frame and correspondent fields), not from a verified Assignment Center pull. I have not fabricated reel/frame numbers or correspondent names — where the task asks for them, I mark them not retrieved. Verification link: https://assignment.uspto.gov/patent/index.html (search 11402387 or 16/616,831).

Cross-reference note: this resolves uncertainty #1 flagged in the earlier summary — I located the literal claim set during this search (independent claims 1 and 14; 14 claims total; claim 1 recites "B ion glycosidic fragment or C ion glycosidic fragment," "inferable constituent is a monosaccharide," and "up to four branches"). That does not change the ownership analysis.

Inventors

Patterns to flag:

• No inventor-departure signal. The one-year-window red flag (all inventors exiting the original assignee within 12 months of filing, a predictor of portfolio fire-sales) is not present — there was only ever one named inventor, and the patent remained with the original assignee through the present date.
• Inventorship discrepancy (carried forward, unresolved). The underlying GlycoDeNovo publication lists eight authors (Hong, Sun, Sha, Pu, Khatri, Yu, Tang, Lin), yet the patent names a single inventor. I report the patent record literally and make no representation about correctness of inventorship. This is a substantive-invalidity/provenance issue, not an ownership-chain symptom — but a diligent acquirer would run it down.
• The related 2025 family member (US 2025/0061975 A1 / WO 2023/130045 A3, "…de novo glycan topology reconstruction techniques") traces to the same assignee, Brandeis University, per onscope's owner profile — consistent with a single-institution portfolio, not an inventor-retention/serial-entrepreneur pattern.

Original assignee

Brandeis University (Waltham, Massachusetts).

• Entity type / line of business: private R1 research university; primary business is education and federally funded research, not product manufacturing or licensing-of-claims.
• Did it ship a product embodying the claims? No commercial product. The claimed subject matter is implemented as research software ("GlycoDeNovo"); the NIH grant acknowledgment describes U01 CA221234 as funding "an open-source software suite for processing glycomics and glycoproteomics mass spectral data." That points to freely distributed academic software, not a shipped commercial offering. No evidence of a Brandeis-branded mass-spectrometry product.
• Current status: Operating. No bankruptcy, dissolution, or acquisition. It remains the assignee of record and the licensor in the 2023 confirmatory license.
• Government interest: NIH grants P41 GM104603 and U01 CA221234; the grant of rights language ("The government has certain rights in the invention") plus the 2023 NIH confirmatory license mean the U.S. government holds a march-in/bayh-dole license interest in this patent. Any acquirer inherits that encumbrance.

Assignment timeline

Two recorded events. Neither reel/frame number nor correspondent of record could be retrieved in this session — I will not invent them.

• 2019-11-25 (recorded; execution date not retrieved) — Reel NNNNNN/NNNN — not retrieved

  • Conveyance: Assignment of assignors' interest (per the Google Patents legal-events entry: "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: HONG, PENGYU
  • Assignee: BRANDEIS UNIVERSITY
  • Correspondent: not retrieved — no repeat-correspondent assessment possible for this chain. (This is precisely the field the task flags as the highest-value tell; it is the one I could not confirm. If you need it, it must be pulled from the Assignment Center record itself or via a paid Docket Navigator/Lex Machina pull.)
  • Context: initial inventor-to-institution assignment — ordinary academic onboarding conveyance, executed and recorded ~5 months after the 2018-06-01 PCT filing and ~2.5 years before the 2022-08-02 grant.

• 2023-09-08 (recorded) — Reel NNNNNN/NNNN — not retrieved

  • Conveyance: Confirmatory license
  • Assignor: BRANDEIS UNIVERSITY
  • Assignee: NATIONAL INSTITUTES OF HEALTH (NIH), U.S. DEPT. OF HEALTH AND HUMAN SERVICES (DHHS), U.S. GOVERNMENT
  • Correspondent: not retrieved
  • Context: statutory government-interest recordation — the Bayh-Dole confirmatory license memorializing federal rights flowing from P41 GM104603 / U01 CA221234. Not an ownership transfer, not a privateering step, and not an NPE event. Ownership of the patent remains with Brandeis.

No other records. There is no post-issuance transfer, no security agreement, no merger, no change of name, and no release in the record. Ownership has been continuous with Brandeis University since 2019.

Timeline diagram

timeline
    title Ownership of US 11402387
    2017 : Priority applications filed
    2018 : PCT application filed
    2019 : Inventor assigns to Brandeis University
    2022 : Patent issued to Brandeis University
    2023 : NIH confirmatory license recorded

NPE / troll-pattern signals

Aggregate count of strong signals: 0. Weak signals: 0 (one unclear — the correspondent field, unclear only because the data was not retrievable, not because of anything in the record).

Verdict

Insufficient data — and the insufficiency cuts entirely in the direction of "no NPE activity."

Reasoning: the record contains only the original inventor→Brandeis University assignment (recorded 2019-11-25) plus a government confirmatory license to NIH/HHS (recorded 2023-09-08). Under the task's own definition, a chain consisting of only the original assignment is an "Insufficient data" posture — but substantively it is the cleanest possible posture: ownership never left the operating original assignee, there were zero chained LLCs, zero known asserters, zero pre-suit transfers, and no litigation to which this patent is a party. The residual uncertainty is not about what happened (nothing did) but about metadata I could not retrieve — reel/frame numbers and the correspondent of record — which are the fields that would let one confirm the absence of a repeat-player recording attorney. If you need a litigation-grade negative on signal #3, pull the two Assignment Center entries directly (search 11402387 / 16/616,831 at https://assignment.uspto.gov/patent/index.html) and read the correspondent blocks; on current evidence I expect them to show ordinary university technology-transfer counsel, not an NPE agent.

Verification links

• USPTO Patent Assignment Search: https://assignment.uspto.gov/patent/index.html
• USPTO Assignment Center: https://assignmentcenter.uspto.gov/
• Google Patents record with legal events: https://patents.google.com/patent/US11402387/en

Unretrieved items (do not treat as "none")

1. Reel/frame for the 2019-11-25 assignment of assignors' interest (Hong, Pengyu → Brandeis University).
2. Reel/frame for the 2023-09-08 confirmatory license (Brandeis University → NIH/HHS).
3. Correspondent of record on both entries, and the execution dates (as distinct from recording dates) for both.
4. Confirmation of whether the 2019-11-25 instrument also captured the two provisionals (62/513,495; 62/531,229) and the PCT, or only the national-stage application.

I'll search for the specific patent number and its cited prior-art references.

I have the examiner-cited references from the patent's front page. Let me verify bibliographic details and check for any additional cited references.

Scope and method statement

I searched for U.S. Patent No. 11,402,387 specifically (no similar-number substitutions) and cross-checked the results against the authoritative full text you supplied. The complete citation record I can verify for this patent comes from the (56) References Cited block on the printed front page, which I retrieved via the USPTO/Google Patents image-PDF rendering at patentimages.storage.googleapis.com/.../US11402387.pdf and the Google Patents page https://patents.google.com/patent/US11402387.

Three up-front caveats, carried forward from the prior sections:

1. The literal claim set is still not verified. The rendering I was given contains description/abstract/figures but no claims column. All claim mapping below is anchored to the four SUMMARY aspects already established (method-broad, method-with-ranking, apparatus, computer-implemented method). I explicitly cannot state which numbered claims are implicated.
2. A reference cited on the face of a patent is not the same as an anticipating reference. Nearly all of the §102 candidates below were cleared during prosecution — the patent issued. I therefore distinguish "cited" from "potentially anticipatory," and I flag where a reference is better characterized as §103 art.
3. OCR artifacts in the printed front page. I reproduce identifiers exactly as OCR'd and flag suspected errors rather than silently correcting them, per the operating rule.

USPTO record confirmation for 11402387

Flag: the OCR renders the law firm as "Quartes & Brady LLP." The well-known Chicago firm is Quarles & Brady LLP, and I am not correcting the OCR in the record above — I am flagging it as a probable OCR artifact.

(56) U.S. Patent Documents cited

1. US 2011/0137570 A1 — Lapadula et al.

• Full citation: U.S. Patent Application Publication 2011/0137570 A1 (Lapadula et al.), published 2011-06-09 (per face: "6/2011"). Face-page classification notation: C12Q 1/34 / 702/19.
• Description: Pertains to computational determination of oligosaccharide/carbohydrate structure from mass-spectrometry fragmentation data. In the glycomics literature this family is associated with the OSCAR algorithm — "Congruent strategies for carbohydrate sequencing. 3. OSCAR: an algorithm for assigning oligigosaccharide topology from MSⁿ data," Anal. Chem. 2005, 77, 6271–6279 (that journal article is separately cited in the '387 record through the related-literature chain).
• Anticipation analysis: This is the single most on-point U.S. patent document in the cited set, because it is directed to assigning oligosaccharide topology from MSⁿ data — i.e., the same problem and the same general input/output as Category A (the broad method aspect: acquire MS → identify fragment/composition ions → reconstruct topology). It is the reference an examiner would most plausibly cite alone under 35 U.S.C. § 102 against Category A. However, on its face it does not appear to disclose the specific claimed mechanism — appending fragment ions to an "inferable constituent" to build a candidate topology building block, storing it in a candidate pool conditioned on a first user-defined mass tolerance, then combining building blocks within a second user-defined mass tolerance. That mechanism is the novelty argument for Categories B and D. Best fit: §102 against Category A; §103 against Categories B/D.

2. US 2015/0340213 A1 — Belov

• Full citation: U.S. Patent Application Publication 2015/0340213 A1 (Belov), published 2015-11-26 (per face: "11/2015").
• Description: A mass-spectrometry instrumentation disclosure (Belov is associated with mass-analyzer/ion-mobility instrument development). The face page gives no further identification.
• Anticipation analysis: Relevant principally to Category C (apparatus) — the inlet/ion-source/mass-analyzer/detector architecture is conventional and would be disclosed by any number of instrument references, including this one. It would not anticipate the controller-programming limitations (the GlycoDeNovo reconstruction logic), which is the only potentially novel part of Category C. Best fit: §102 as to the hardware elements of Category C; §103 as to the whole claim.
• Confidence: Medium. I could not retrieve the full text of US 2015/0340213 A1 in this session.

(56) Foreign Patent Documents cited

3. CN 106404883 A

• Full citation: Chinese Patent Application Publication CN 106404883 A, published 2017-02 (per face: "2/2017").
• Description: Not verified in this session. The publication date falls eight months before the '387 priority date, so it is within the §102(a)(1)/(a)(2) window if its subject matter is on point.
• Anticipation analysis: Cannot assess without the document. I will not speculate as to its content. It would need to disclose the append-to-inferable-constituent/candidate-pool mechanism to anticipate Categories B/D, or a full topology-reconstruction-from-MS method to anticipate Category A. Best fit: unknown — flagged for retrieval.

4. WO 2009/154964 A2 (rendered on the face as "WO 2099154964 A2")

• Full citation: PCT publication, December 2009 (per face: "12/2009").
• Flag: the face-page OCR reads "WO 2099154964 A2," which is malformed — an eleven-digit serial after "WO" is not a valid PCT publication number. I am not auto-correcting it. The most plausible intended identifier is WO 2009/154964 A2, but I have not verified this against the WIPO/Patentscope record and I state it only as a lead.
• Description / anticipation analysis: Not verified. Cannot assess. Note that a 2009 publication predates the priority date by roughly eight years, so if its subject matter were on point it would be squarely §102 prior art — which makes it worth retrieving.

(56) Other Publications (non-patent literature) cited

The front page carries a long NPL list. Where an entry carries the examiner-citation marker on the face, I note it. Several entries have OCR corruption in author/page fields; I flag these rather than correct them.

Consolidated §102 mapping (by claim category)

Reminder: claim categories are the four SUMMARY aspects carried over from the prior sections, not verified claim numbers.

Where I would look next for true §102 art: the two unverified foreign documents — CN 106404883 A (2017-02) and the malformed "WO 2099154964 A2" (2009-12) — plus a full-text pull of US 2015/0340213 A1. A February 2017 Chinese application on glycan MS analysis would fall after the June 2017 priority date and therefore could not anticipate; a December 2009 PCT publication could.

Explicit uncertainties

1. I did not obtain the literal claim set, and I did not obtain the full text of US 2011/0137570 A1, US 2015/0340213 A1, CN 106404883 A, or WO 2009/154964 A2. The §102 mappings above are therefore potential only.
2. The "Kumczaki 2015" bibliographic data is unverified — author surname and page numbers are shown as printed. I confirmed the subject matter (machine-learning de novo glycan sequencing) but not the citation string.
3. The "WO 2099154964 A2" identifier is malformed in the source OCR. I deliberately did not normalize it to WO 2009/154964 A2 in the citation table.
4. Examiner-cited vs. applicant-cited: I identified three face-marked examiner citations (Lapadula, Saar-Tsechansky, Carlsson). The remaining NPL entries appear to be applicant IDS citations by ordering, but the OCR layout is not conclusive, so I do not assert the distinction for those.
5. No USPTO "cited by" (forward-citation) or litigation analysis is included here — the prior sections already establish that no litigation or PTAB proceeding against 11,402,387 was found, and this task is limited to the backward (56) citations.

I'll start by trying to recover the actual prior-art record for this patent (the provided page text only includes prior-art keywords and a prior-art date, not the references themselves), then search for the substantive glycan-MS art that would be combinable.

I now have the actual § 56 "References Cited" list from the patent's face. Let me verify the two U.S. patent documents, the two foreign documents, and the most important non-patent references (especially Kumozaki's machine-learning glycan sequencing paper).

Obviousness Analysis under 35 U.S.C. § 103 — U.S. Patent No. 11,402,387 B2

Prepared: April 26, 2026 · Subject: System and method for determining glycan topology using tandem mass spectra (Brandeis University; inventor Pengyu Hong)

0. Scope, sources, and critical caveats (read first)

0.1 What "the Prior Art section of this page" actually contained. The page text supplied to me contains a "Prior art keywords" field (ion; mass; topology; fragment ions; fragment) and a prior-art date (2017-06-01), but no references. I therefore recovered the patent's actual § 56 record from the granted-patent PDF, which is the authoritative list of art that the examiner cited and considered. That list is reproduced in §1 below. Sources:

• Granted patent PDF (References Cited): http://patentimages.storage.googleapis.com/06/93/3f/a4ce4d61715d8b/[US11402387](/patent/US11402387).pdf
• Google Patents: https://patents.google.com/patent/US11402387/en
• Justia: https://patents.justia.com/patent/11402387

0.2 The claim numbers I analyze are from the pre-grant publication, not verified granted claims. The full text supplied to me does not contain a claims section. I obtained a 20-claim set from the pre-grant publication US 2020/0096518 A1 (https://www.patentsencyclopedia.com/app/20200096518). The a)/b)/c) step-lettering in claims 1, 15 and 20 bears the fingerprints of a post-filing amendment, so the granted claims may differ. Everything below is conditioned on that set being identical or substantially identical to the granted set. This is my single largest uncertainty.

0.3 Identifiers are interpreted literally; OCR artifacts are flagged, not corrected. The § 56 list as OCR'd contains "WO 2099154964," "Kumczaki," and "Bocker … 976-860." On the face of the record these are what the patent prints; substantively they correspond to WO 2009/154964 A2, Kumozaki, and Böcker … 976–986. I flag the discrepancy rather than silently correcting it. If the printed number is wrong, that is itself a § 112/§ 132 citation-accuracy issue, not a § 103 issue.

1. The § 56 prior-art record (the art to be combined)

Key procedural observation. The examiner's own "Prior art keywords" (ion, mass, topology, fragment ions, fragment) map one-to-one onto the claimed elements, and the § 56 list is essentially a complete survey of the de novo glycan-sequencing field circa 2000–2015. That is the single most important fact for a § 103 analysis: virtually every element of claim 1 exists somewhere in this list, and the applicant's own specification concedes the field's state of the art (see §3.1).

2. Level of ordinary skill (POSITA)

A POSITA here is a person with an M.S. or Ph.D. in chemistry, bioinformatics, computer science, or a related discipline with 2+ years' experience in mass-spectrometric glycan analysis and computational algorithm design, or equivalent. The POSITA is comfortable with (i) Domon–Costello fragmentation nomenclature (B/C/Y/Z, A/X), (ii) graph/tree dynamic programming, (iii) machine-learning classifiers (decision trees, boosting) over spectral feature vectors, and (iv) instrument hardware (quadrupole filters, collision cells, FT-ICR/Orbitrap). All of the asserted § 56 art is squarely within this person's field. This matters: the narrower the POSITA's field and the more the references share it, the stronger the motivation-to-combine finding (In re Keller, 642 F.2d 413 (CCPA 1981); MPEP 2144.01).

3. The independent claims

3.1 Claim 1 (method: acquire → identify via building blocks in a candidate pool → reconstruct)

Prima facie case. Claims 1 and 20 read on the combined teachings of Dong (2015) + Böcker (2011), optionally with Tang (2005) and Lapadula (US 2011/0137570 / WO 2009/154964).

Motivation to combine (KSR, 550 U.S. 398, 417–421 (2007); MPEP 2143):

1. Common field and common problem. All four references address the same problem: reconstructing a glycan's topology from MS/MS without a database. The patent's own Background says so verbatim: "Currently, there is a need for a reconstruction technique that can accurately characterize the structure of both large macromolecules and small molecules with reduced computational complexity, and through the use of a method that does not rely on a database of known structures." That sentence is an admission of the pre-existing need that motivates the combination (MPEP 2143; In re Fritch, 972 F.2d 1260 (Fed. Cir. 1992)).
2. Known problem, known solutions, predictable improvement. Dong expressly identifies the "repetitive peak counting" defect of tree-based methods (which the patent at ¶0037 calls "a long felt, but unresolved need in the art") and solves it with its DAG six-tuple + union operation. Böcker then shows the tree-counting problem is polynomially solvable. Combining Dong's reusable-substructure strategy with Böcker's mass-bounded exact enumeration is nothing more than applying a known technique to a known method to yield a predictable result — a textbook KSR rationale.
3. Design incentive / finite number of identified solutions. The art had a small, finite set of approaches (catalog-library; relationship tree; PRM/DP; DAG/block; FPT tree-matching; Lagrangian-relaxation ML). When the field converges on a handful of known alternatives, KSR makes combination "obvious to try."
4. Mass tolerance is a routine, recognized parameter. Every one of these references necessarily operates within a mass-accuracy window (Böcker's fixed-parameter bound; Dong's "logical constraints"; GlycoMod's composition mapping). Selecting 0.02 Da / 0.005 Da is the optimization of a recognized parameter — obvious under In re Boesch, 617 F.2d 272 (CCPA 1980), absent unexpected results.
5. Reasonable expectation of success. All references are algorithmic/software; implementing a candidate-pool data structure over lighter-mass building blocks is a mechanical programming step with a predictable outcome — In re O'Farrell, 853 F.2d 894 (Fed. Cir. 1988).

The applicant's own characterization helps the challenger. The patent concedes GLYCH (Tang) has the same O(N^(H+1)) complexity as the claimed PeakInterpreter. A claimed advantage that the applicant admits the prior art already achieves cannot support patentability under § 103.

3.2 Claim 20 (computer-implemented variant)

Identical to claim 1, with the identifying/reconstructing steps performed "using the computer system." Lapadula, Dong, Böcker, Tang, Kumozaki are all computer-implemented methods; running a known algorithm on a general-purpose computer (or distributing its steps across a computer and a mass spectrometer) is the epitome of an obvious, predictable implementation (In re Venner, 262 F.2d 91 (CCPA 1958); MPEP 2144.04). Claim 20 rises and falls with claim 1.

3.3 Claim 15 (mass spectrometry unit, apparatus)

Motivation to combine Belov with the glycan-sequencing software art: a POSITA building a glycan sequencer would be motivated to use an instrument already configured to (i) isolate a precursor, (ii) fragment it to monomer subunit ions, (iii) mass-select the subunit ions, (iv) fragment them further to MS³, and (v) measure at high mass accuracy — all of which Belov discloses and all of which the patent's own specification demands (FT-ICR, <2 ppm accuracy, EED/MSⁿ). The patent's specification itself concedes that any MS capable of ionizing and separating by m/z will do (AMS, GC-MS, LC-MS, MALDI-TOF, SELDI-TOF, tandem MS, TIMS, SSMS), which is fatal to a nonobviousness position built on hardware. Note the applicants did not dispute this in prosecution insofar as the record shows, and the examiner originally cited Belov.

3.4 Claims 2 and 10–13 (ranking by a machine-learning topology score)

This is the only genuinely contestable cluster, and it is not, in my view, non-obvious over the § 56 art:

Motivation to combine Kumozaki with Dong/Böcker: the problem is stated in the art itself — Tang's scoring "may repeatedly use peaks… which should be avoided because it favors linear structures over branched ones" (Böcker, citing Tang); Shan shows the problem is NP-hard; Kumozaki applies ML precisely to fix scoring. Combining a known ML scoring model with a known candidate-generation algorithm is the application of a known technique to a known method, and the patent's own stated benefit ("automatically learn fragmentation patterns from real-world tandem MS data") is the expected result of training a classifier — not an unexpected one.

The strongest § 103 position on claims 10–13 is Kumozaki alone or Kumozaki + Saar-Tsechansky/Carlsson, because the examiner already cited both of the latter two — i.e., the examiner had the generic ML-classifier teachings on the record, and the only specific hook the applicant added over Kumozaki (a 105 Da window, ±1 labels) is a routine design choice.

Caveat — where the rejection is weakest: if the granted claim 10 or 11 requires scoring based specifically on the non-reducing-end B/C-ion classification (i.e., the IonClassifier distinguishing B/C from Y/Z/O ions and using that as the candidate score), a challenger should show Kumozaki's learned features are applied to ion-type discrimination and not merely to structural-element scoring. I could not retrieve Kumozaki's full text in this session, and I flag that as a genuine gap (see §6).

4. Recommended combinations (prosecution-grade rejections)

Rejection A — independent claims 1, 20 (and 15):
Dong (2015) in view of Böcker (2011), further in view of Lapadula (US 2011/0137570 A1 / WO 2009/154964 A2) and Tang (2005).

• Dong: reusable "blocks," candidate pool of top-N substructures, mass-constrained constraints, explicit anti-peak-double-counting.
• Böcker: exact, polynomial tree reconstruction from MS alone, mass-bounded candidate sets.
• Lapadula: "grow the candidate structures," substructure-to-fragmentation-value mapping, threshold-of-acceptability scoring.
• Tang: PRM building blocks anchored to a residual constituent (the claimed "inferable constituent").

Rejection B — claims 2, 10, 11, 12 (ML ranking): Rejection A further in view of Kumozaki (2015), and, for the generic classifier framework, Saar-Tsechansky (2007) and Carlsson (2012).

Rejection C — claim 15 (apparatus): Belov (US 2015/0340213) in view of Dong/Böcker/Tang/Lapadula. Belov supplies every hardware element; the cited algorithmic references supply the controller programming.

Rejection D — claim 13 (105 Da window): Rejection B + In re Boesch (parameter optimization), or Kumozaki's own feature windows as an alternative.

5. Anticipated applicant rebuttals and how they fare

1. "The references teach away / are non-analogous." Weak. All are glycan MS/MS topology-reconstruction references; the problem and field are identical. Belov is analogous because the apparatus claim is directed to the same function (generating and mass-analyzing subunit/fragment ions), even though Belov's exemplified analytes are protein complexes (In re Bigio, 381 F.3d 1320 (Fed. Cir. 2004); Sanofi-Synthelabo v. Apotex, 550 F.3d 1075 (Fed. Cir. 2008)).
2. "Delayed reconstruction / avoiding the Peak Assignment Problem is unexpected." Not unexpected. Böcker expressly solved the Peak Assignment Problem (fixed-parameter tractability) and Dong expressly solved repeated peak counting two years before the provisional. The specification's assertion at ¶0037 that this was "a long felt, but unresolved need in the art" is contradicted by the very references the examiner cited — Shan (2008) frames and attacks the NP-hardness; Böcker (2011) solves it; Dong (2015) reports a working solution. Attorney argument in the specification is not objective evidence of nonobviousness (In re GPAC Inc., 57 F.3d 1573 (Fed. Cir. 1995)).
3. "Complexity is polynomial rather than exponential." The applicant's own specification concedes GLYCH (Tang) already had the identical O(N^(H+1)) complexity. Where the advantage is admitted in the art, there is no nonobviousness to rest on.
4. "The experimental rank improvements (e.g., ¹⁸O-Man9 from 205th to 1st) show unexpected results." Not on this record. (i) The result is a ranking improvement, i.e., a matter of degree, and KSR permits "superior results" that are merely differences in degree where the result was predictable. (ii) No evidence in the provided text establishes a statistically significant, and unexpected improvement over the closest art (Kumozaki) — the comparison is against the applicant's own SPN baseline, not against the prior art (In re Baxter Travenol Labs., 952 F.2d 388 (Fed. Cir. 1991)). (iii) The 2017 ASMS abstract/paper is the applicant's own unpublished-work; self-comparisons do not establish nonobviousness over the § 56 art.
5. "The examiner allowed the claims over this art." Allowance is not a legal conclusion binding in a validity analysis; it may reflect an amendment (the a)/b)/c) step-lettering in claims 1, 15 and 20 strongly suggests one) rather than a substantive finding of nonobviousness over Dong + Böcker + Kumozaki. Absent the prosecution history, this argument should be probed before relying on it.

6. Secondary considerations and § 102 observations

6.1 No objective indicia are on the record I can see. No commercial embodiment, licensing, copying, industry praise, or nexus evidence appears in the provided text. The only "long-felt need" assertion is the specification's argument at ¶0037, rebutted above.

6.2 § 102 note (not asked, but material to a § 103 posture). Two of the examiner's "Other Publications" are the applicant's own contemporaneous disclosures: the ASMS abstract (Aug. 7, 2017) and the GlycoDeNovo JASMS paper (Hong et al., Nov. 2017, 28(11):2288–2301, PMC5647224). Both post-date the 2017-06-01 provisional, so they are presumptively exempt under § 102(b)(1)(A) as the inventor's own grace-period disclosures for subject matter supported by the provisionals. But if the granted claims recite the IonClassifier machine-learning limitations, and if the June 1, 2017 provisional did not support them (the second provisional is dated 2017-07-11), the argument space narrows, and the applicant's own ASMS/JASMS disclosures become potential § 102(a)(1) art to be sworn behind. This is worth checking against the provisional texts.

6.3 Claim-drafting anomalies to verify literally (they affect the analysis, so I do not correct them):

• Claim 13 in the pre-grant set reads: "The method of claim 9, wherein the mass-difference window is approximately 105 Da" — claim 9 is directed to branches, not to a window. Either a dependency error or a renumbering between publication and grant.
• Claim 19 ("The method of claim 1, wherein the inferable constituent comprises a monosaccharide") duplicates claim 6.
• Claim 1 recites "determining a topology for a molecular" (not "molecule").
• Claim 17 recites "The mass spectrometer of claim 15" (not "mass spectrometry unit").
  These are inconsistent-claim issues under 35 U.S.C. § 112(b), independent of § 103.

7. Bottom line

Strongest challenger case: a three-reference rejection (Dong + Böcker + Kumozaki) with Belov for the apparatus claims and Lapadula/Tang for the building-block/pool limitations.
Strongest patentee case: (i) the claim set as granted may contain narrowing language I have not seen, and (ii) Kumozaki's precise feature set must be compared element-by-element to granted claim 10/11.

8. Confidence and explicit gaps

• High confidence in the identity, dates, and substance of the § 56 non-patent references (Böcker 2011; Dong 2015; Tang 2005; Gaucher 2000; Shan 2008; Kumozaki 2015; Ethier 2002/2003; Lohmann 2004; Cooper 2001; Tseng 1999), verified against the granted patent's face and independent abstracts at PubMed/IEEE/ACM.
• High confidence in Lapadula's and Belov's disclosures (verified from their own published texts).
• Medium confidence in the claim numbering/content because I am relying on the pre-grant publication's 20-claim set, not the granted claims.
• Known gaps I did not close: (i) the full text of Kumozaki 2015 (abstract and secondary characterizations only); (ii) CN 106404883 A; (iii) the prosecution history (which would reveal exactly which reference was overcome and how, and whether the a)/b)/c) step-lettering reflects an amendment); and (iv) the provisional specifications for the § 102(b)(1)(A) grace-period question. I did not fabricate substance for those items; where I could not verify, I said so.
• No litigation exists involving this patent (consistent with the earlier litigation section), so there is no validity challenge or IPR record to work from — this is an a priori obviousness assessment.