US 7537897

I'll search for authoritative information on this specific patent number and check for any CAFC litigation.

Let me search specifically for CAFC 2026 dockets and current litigation status.

Search results summary — US 7,537,897

Scope note / uncertainty: I ran the requested targeted searches (USPTO/PTAB records, Docket Alarm, IPVerse, Federal Circuit opinion listings, and general web). I did not find any CAFC docket in 2026 in which 7,537,897 itself is the patent-in-suit. Absence of a hit in these third-party aggregators is not conclusive proof no litigation exists, but nothing found points to a 2026 appeal for this number. Details and the near-misses I did find are set out at the end.

Bibliographic data (as published on the granted patent)

Priority is claimed from six U.S. provisionals: 60/761,578 (23 Jan 2006); 60/775,098 (21 Feb 2006); 60/777,661 (27 Feb 2006); 60/779,540 (6 Mar 2006); 60/791,561 (12 Apr 2006); and 60/824,456 (4 Sep 2006).

Ownership history: filed by Population Genetics Technologies Ltd → assigned to Compass Genetics, LLC (Mar 14, 2007) → back to Population Genetics Technologies Ltd (Apr 28, 2008) → assigned Oct. 20, 2017 to Agilent Technologies Singapore Holdings Pte. Ltd. → then to Agilent Technologies, Inc. That is why different sources list either Population Genetics Technologies or Agilent as assignee; both are correct as of different dates.

Abstract (verbatim)

"The invention provides methods and compositions for counting molecules in a sample, wherein each molecule is labeled with a unique oligonucleotide tag. Such tags are amplified and identified rather than the molecules themselves; that is, the problem of counting molecules is converted into the problem of counting tags. In one aspect of the invention, molecules to be counted are labeled by sampling. That is, conjugates are formed between the molecules to be counted and oligonucleotide tags of a very large set, or repertoire. After conjugation, a sample of conjugates is taken that is sufficiently small so that substantially every molecule has a unique oligonucleotide tag. Counting of different tags may be accomplished in a variety of ways. In one aspect, different tags may be counted by carrying out a series of sorting steps to generate successively less complex mixtures in which tags are enumerated using length-encoded 'metric' tags. In another aspect, different tags may be counted by directly sequencing a sample of tags using any one of several different sequencing methodologies."

Plain-language overview of each independent claim

There are five independent claims: 1, 8, 12, 17 and 20 (claims 2–7 depend from 1, 9–11 from 8, 13–16 from 12, 18–19 from 17).

Claim 1 — Counting molecules by repeatedly "sorting" the tags.
Build conjugates in which essentially every distinct molecule in the sample carries a different oligonucleotide tag, where each tag is a concatenation of subunits ("words") drawn from a defined set of 2 to 6 members, each subunit occupying a position. Then (b) divide the tags into aliquots by sorting on the identity of the subunit at a chosen position (first position, then successive positions), and (c) keep repeating the division on at least one aliquot until an aliquot contains no tags left that can be subdivided. The molecule count is derived from how many times the sorting step had to be applied.

Claim 8 — Counting target polynucleotides using two functionally distinct tag types.
Label each target polynucleotide "by sampling" so essentially each one gets a unique tag composed of a sorting tag and an identification tag. Successively sort the tags by their sorting tags to produce one or more separate, simpler mixtures; then determine how many different tags are in at least one of those mixtures using the identification tags. The target count is read out from the number of sorting rounds combined with the number of different tags found in the sorted mixture(s). (Dependent claims add that the sorting tag is a binary tag — ideally a dinucleotide concatenate — and that the identification tag is a "metric" tag, with far more binary tags than metric tags.)

Claim 12 — Counting by nuclease-resistant selected probes.
For each target polynucleotide, supply a plurality of nucleic acid probes each bearing a different tag. Combine probes with targets so that essentially every target associates with a probe to form a "selected probe" that is resistant to a nuclease activity; the probe pool is made much larger than the number of targets so essentially every selected probe carries a unique tag. Isolate the selected probes by treating the mixture with a nuclease, then sequence the tags in a sample of the isolated probes and count the different tags to obtain the target number. (Dependents add restriction-fragment targets with unique overhangs, a circularizing adaptor that forms a dsDNA circle on ligation, type IIs — including "double cleavage" type IIs — digestion, and concatenation of tags before sequencing.)

Claim 17 — Estimate target number by sequencing amplified tags.
Label each target polynucleotide by sampling so essentially each has a unique tag; amplify the tags; then determine how many different tags are present in a sample of the amplified tags by sequencing them, thereby estimating the number of targets. (Dependents add emulsion PCR and pyrosequencing/Sanger/ligation-based sequencing.)

Claim 20 — Estimate target number by counting length-encoded bands (no sequencing).
Label each target polynucleotide by sampling so essentially each has a unique metric tag; amplify the metric tags; separate the amplified tags to produce a separation profile of distinct bands; and count the distinct bands to estimate the number of targets in the mixture.

Litigation / PTAB status found (and near-misses)

• No 2026 CAFC appeal naming 7,537,897 as the patent-in-suit was located.
• PTAB IPR2023-00955 (Parse Biosciences, Inc. v. 10x Genomics, Inc.) — the patent at issue is 10,240,197, not 7,537,897. US 7,537,897 appears there only as Petitioner's Exhibit 1058 (i.e., cited as background/prior-art-type evidence). That IPR reached a final written decision Jan. 31, 2025, and the appeal is 10x Genomics, Inc. v. Parse Biosciences, Inc., Nos. 25-1199, 25-1618/-1619 (Fed. Cir.), decided by nonprecedential opinion Aug. 19, 2026, affirming the Board. The patents actually on appeal were 10,155,981; 10,240,197; and 10,697,013 — again not 7,537,897. (Compare: https://www.cafc.uscourts.gov/opinions-orders/25-1199.OPINION.8-19-2026_2741460.pdf and https://fedcircuitblog.com/2026/08/19/opinions-orders-august-19-2026/)
• 10x Genomics, Inc. v. Celsee, Inc., No. 1:19-cv-00862 (D. Del.) — US007537897B2 appears as an attachment/exhibit to a filing (Doc. 104-3, filed May 22, 2020), not as the asserted patent.
• Several later patents (e.g., Guardant/Stanford/Becton Dickinson families) cite 7,537,897 as prior art or as incorporated background.

Bottom line: Bibliographic data, abstract, and all independent claims are corroborated by multiple independent sources and by the authoritative full text. The one genuinely uncertain item is whether any 2026 litigation activity exists that is not indexed by the aggregators searched — I found none, and I am flagging that as a search-coverage limitation rather than a definitive clearance.

I'll search for litigation involving US patent 7537897.

Let me search more targeted litigation sources.

Let me search Google Patents' litigation data and specific case databases.

Summary of findings for U.S. Patent No. 7,537,897

Bottom line: I found no record of any litigation (district court or Federal Circuit appeal) in which U.S. Patent No. 7,537,897 ("Molecular counting," Brenner & Macevicz; filed Jan. 22, 2007; granted May 26, 2009; original assignee Population Genetics Technologies Ltd.; now Agilent Technologies, Inc.) was asserted as a patent-in-suit. I cannot confirm any plaintiff–defendant case involving this patent as an asserted claim.

What I was able to verify

Critical distinction (please note)

The IPR2023-00955 reference is not litigation involving the '897 patent. In that proceeding Parse Biosciences used US 7,537,897 as an exhibit (prior-art/background support) in a petition challenging a different patent — the docket snippet indexes it against US 10,240,197, not '897. No party there is litigating the '897 patent.

Identifiers I deliberately did not treat as this patent

• FR 7537897 — a 1975 French application number (FR2294481B1, a 1974 German priority pressure-regulator valve). Unrelated.
• US 7,553,818 (appears in a drug-patent-watch result) — a different patent. Ignored per instruction not to substitute similar numbers.
• US 7,545,473 / 8,476,018 / 8,718,609 etc. — separate Brenner/PGT-family or unrelated patents appearing only as citations.

Limitations / caveats (stated explicitly)

1. I was unable to complete a fully exhaustive check of the primary litigation databases. My attempt to reach additional targeted queries (including additional Unified Patents portal / PACER-style lookups) was cut off before completion, so this is not a 100% complete census.
2. The Unified Patents litigation caselist, Docket Alarm, Justia Dockets, CourtListener, and RPX were not each individually queried to exhaustion for this specific number. My conclusion rests on the searches actually performed (Google Patents family/litigation data, Docket Alarm PTAB exhibit index, general web/case-law searches, and searches combining the patent owner names).
3. Absence of a hit does not prove no litigation exists — only that I could not find any.

Recommended verification steps

To confirm with high confidence, I would run the patent number through:

• Unified Patents litigation portal (portal.unifiedpatents.com/litigation) — search "7537897"
• Docket Alarm / Justia Dockets / CourtListener RECAP — full-text complaint and docket search for "7,537,897"
• PACER and the CAFC docket for any appeal involving the patent
• USPTO PTAB (IPR/PGR) docket search, to capture any challenge to '897 (not found in my search)
• RPX Insight litigation search

Conclusion: Based on available information, there is no known litigation asserting or challenging U.S. Patent No. 7,537,897. Its only appearance in a contested proceeding is as a prior-art exhibit in Parse Biosciences' IPR2023-00955 petition, which is not a case about the '897 patent. I recommend the verification steps above before relying on this as definitive.

I'll verify the structured block with targeted searches before finalizing.

Proceedings overview

Zero AIA trial proceedings are on file for US 7,537,897 — 0 active, 0 claims invalidated, 0 claims sustained, 0 settled, and 0 institution-denied. The patent has never been subjected to an IPR, PGR, or CBM challenge at the PTAB, so the defensive posture is not "the patent has been hardened by surviving IPRs" and not "claims were canceled"; it is the third and least comfortable case — the claims are entirely untested at the Board, meaning there is no FWD to point to, no claim-cancellation to rely on, and no petitioner estoppel to exploit, but also no PTAB litigation history that a patent owner can wave at the district court as a validity endorsement.

This finding rests on two independent sources:

1. Canonical structured source (controlling): the USPTO Open Data Portal (ODP) "PTAB proceedings on file" block supplied with this analysis returns no AIA trial proceedings for this patent as of the most recent ingest.
2. Independent corroboration (web): targeted searches of PTAB docket aggregators, IPR-exhibit indices, Federal Circuit opinion listings, and the IPVerse company-level PTAB dataset surfaced no challenge proceeding naming 7,537,897 as the challenged patent. IPVerse's competitive-analysis table for Population Genetics (last updated 2026-09-10) shows that company with 8 oppositions faced but zero PTAB cases faced and zero PTAB cases filed.

Two structural facts narrow the realistic challenge space and are worth stating for a defendant up front:

• PGR is unavailable. The '897 patent has a priority date of 2006-01-23 and was filed 2007-01-22 — pre-AIA / first-to-invent. PGR under 35 U.S.C. § 321 applies only to first-inventor-to-file patents. That door closed before the patent issued.
• CBM is unavailable. The claims are directed to molecular counting / nucleic acid analytics, not to "a method or corresponding apparatus for performing data processing or other operations used in the practice, administration, or management of a financial product or service." No covered-business-method hook.
• IPR is the only realistic AIA vehicle, and it has never been used.

(No proceeding) — the one PTAB docket entry that touches this patent number, and why it is not a proceeding on '897

There is exactly one PTAB record in which the string "7,537,897" appears, and it is not a challenge to this patent.

• Apparent proceeding: IPR2023-00955, Parse Biosciences, Inc. v. 10x Genomics, Inc.
• Type: Inter Partes Review (but of a different patent — the challenged patent is US 10,240,197)
• Filed (of the exhibit): 2023-05-25
• Status of '897 in that matter: Petitioner's Exhibit 1058 — i.e., the granted '897 patent was submitted as prior-art / background evidence to support Parse's challenge to '897's descendant patent 10,240,197. The '897 patent was neither challenged nor construed in that IPR.
• Judge panel: not applicable to '897 (panel decided the '197 patent, not '897)
• Petition grounds: not applicable to '897; no § 102/§ 103/§ 112 ground was ever run against '897 in this matter
• Institution decision: not applicable to '897
• Final Written Decision: not applicable to '897. For completeness on the other patent: the Board's FWD in IPR2023-00955 issued 2025-01-31, and the Federal Circuit affirmed by nonprecedential opinion in 10x Genomics, Inc. v. Parse Biosciences, Inc., Nos. 25-1199, 25-1618/-1619 (Fed. Cir. 2026-08-19). The patents actually on appeal were 10,155,981; 10,240,197; and 10,697,013 — not 7,537,897.
• Settlement / termination: N/A to '897
• Appeal: N/A to '897
• Defensive value: Essentially none as to validity — but high as to opposing-counsel hygiene. Expect a plaintiff or a defense team to conflate the Parse line of cases with this patent number, because the '897 document sits in the exhibit list and the "Brenner patents" family branding is shared. If you see a brief citing IPR2023-00955 as though it adjudicated 7,537,897, that is a misstatement of the record. Source the exhibit for yourself: https://www.docketalarm.com/cases/PTAB/IPR2023-00955/Parse_Biosciences_Inc/05-25-2023-Petitioner/Exhibit-1058-US_7,537,897_B2/

The same document also surfaces as an exhibit in district court — 10x Genomics, Inc. v. Celsee, Inc., No. 1:19-cv-00862 (D. Del.), Doc. 104-3, filed 2020-05-22 — again as a supporting attachment, not an asserted patent. In that litigation the "Brenner" patents asserted were U.S. 10,155,981; 10,697,013; and 10,240,197, and Celsee's invalidity motions (e.g. the § 112 motion at D.I. 263) were directed at those, with '897 cited by 10x as an incorporated-by-reference disclosure bearing on counting polynucleotides. That is the opposite posture from being a patent-in-suit.

Non-running PTAB-adjacent items to rule out explicitly (so you don't chase them):

• No ex parte or inter partes reexamination naming 7,537,897 was surfaced. The 2022-04-11 inter partes reexamination record that appears in one search hit (Control No. 95/002,170) concerns US 7,897,080 (In re Yang), an unrelated patent — do not carry that control number over.
• A Certificate of Correction exists on '897 (noted by uspto.report; content not retrieved — treat as unverified). A certificate of correction is a § 255 clerical/minor-error mechanism, not a PTAB proceeding; it does not narrow claims on validity grounds and creates no estoppel. Verify its text before relying on any characterization of the issued claims. https://uspto.report/patent/grant/7,537,897
• Term: anticipated expiration 2027-01-22. As of 2026-09-28 that is roughly 4 months of remaining term.

Strategic summary

Claim status: every claim of 7,537,897 is UNTESTED. There is no IPR certificate, no FWD, and no reexamination certificate of any kind for this patent. All 20 claims — including independent claims 1, 8, 12, 17 and 20 and dependents 2–7, 9–11, 13–16, and 18–19 — remain exactly as granted, subject only to whatever the unverified Certificate of Correction changed (most likely clerical). You cannot say "claim 1 is dead" and you cannot say "claims 1–5 have been canceled." Neither is true. Practically, this means a demand letter citing any of the five independent claims is citing live, unchallenged, never-PTAB-reviewed claims.

Estoppel landscape: empty — and that cuts in the defendant's favor. Because no IPR ever reached a final written decision, 35 U.S.C. § 315(e)(2) estoppel has never attached to anyone. There is no petitioner, no privy, and no "ground raised or reasonably could have been raised" that is closed off. A defendant today can file a fresh IPR on any § 102/§ 103 ground based on patents and printed publications, with no General Plastic "follow-on petition" baggage from a prior petitioner (though Genera Plastic is panel-discretionary, not a statutory bar, and the Board does discount petitions filed after an earlier one on the same patent). The art-of-record list provides a useful starting inventory for a § 102/§ 103 attack, including the references the examiner already considered — Daser et al., "Interrogation of genomes by molecular copy-number counting (MCC)," Nature Methods 3(6):447-453 (2006); Hirano and Fukami, "A novel method for DNA molecular counting," Nucleic Acids Symposium Series No. 44:157-158 (2002); Jarvius et al., Nature Methods 3(9):725-727 (2006); Vogelstein et al., "Digital PCR," PNAS 96:9236-9241 (1999); Brenner et al., U.S. Pat. No. 5,846,719; and WO 2005/080604. None of those was ever tested in a contested AIA proceeding for this patent.

Pattern signals: none of the usual tells are present. No serial petitioner. No repeat-IPR pattern. No patent-owner win at the Board to brag about, and no Board loss to hide. No indication that a defensive aggregator such as Unified Patents ever targeted this patent — the IPVerse dataset shows zero PTAB cases on the Population Genetics side of the ledger. The near-miss pattern is instead family confusion: the litigation and PTAB activity in this space (10x v. Celsee; Parse v. 10x; the 2025-01-31 FWD; the 2026-08-19 Federal Circuit affirmance) all run against the later Brenner-family patents 10,155,981, 10,240,197, and 10,697,013, with '897 appearing only as an exhibit or an incorporated-by-reference parent. If someone tells you "the Brenner molecular-counting patents have been through IPR," that is true of those three — not of 7,537,897.

One further wrinkle a defendant should check before filing. The PTAB, via the Director's 2025-10-16 memorandum and the 2025-10-15 proposed rules (comment period closed 2025-11-17), moved toward barring IPR against any patent that had already survived a validity challenge in district court or at the PTAB, and toward requiring § 102/§ 103 stipulations in parallel venues. Whatever the final status of those rules as of 2026-09-28 (I am not certain whether they were finalized; verify), that bar would not apply to '897, because '897 has never survived a validity challenge in any forum. The absence of PTAB activity here is therefore permissive rather than preclusive.

Recommended next steps

1. Do not accept any assertion that an IPR decided anything about 7,537,897. There is no FWD for this patent to cite, quote, or attach. If opposing counsel points to IPR2023-00955, make them identify the challenged patent — it is 10,240,197, and '897 is Exhibit 1058. Verify on PTAB E2E: https://ptab.uspto.gov/ and cross-check the exhibit at the Docket Alarm link above.
2. Refresh the ODP/PTAB check immediately before you rely on this memo. The canonical block is the USPTO ODP ingest, and ODP indexing can lag a newly filed petition by weeks. Re-run the patent number through PTAB E2E and PatentCenter (https://patentcenter.uspto.gov/) and confirm there is no Notice of Filing Date Accorded filed in the last 60–90 days.
3. Pull the Certificate of Correction text. uspto.report flags one as existing for this grant; its content was not retrieved in this analysis. It could affect the literal claim language you must map. Order the corrected grant and compare claims 1, 8, 12, 17 and 20 line-by-line against the pre-grant publication US 2007/0172873 A1.
4. Budget for the clock, not for the Board. With anticipated expiration 2027-01-22, an IPR filed now would very likely not reach a final written decision until after the patent has expired, converting the exercise into a claim-scope/§ 287 damages fight rather than a term-clearing one. If your real exposure is past damages, the more efficient levers may be district-court § 102/§ 103/§ 112 invalidity (as Celsee attempted against the sibling patents), claim-construction/§ 101 arguments, and § 315(b) timing discipline if you have already been served — remember you have one year from service of a complaint alleging infringement to file an IPR, and that is a hard statutory bar.
5. Treat the '897 specification as a two-way street. It has already been used as prior art against a later patent (Parse, Ex. 1058), and it is expressly incorporated by reference into later Brenner-family patents. That makes it valuable to a defendant who wants to argue that the later claims are not supported or are anticipated — but it also means the '897 disclosure has been on the public record since 2007-07-26 (pre-grant pub.) and cannot be characterized as confined art.

Bottom line for a defendant: the absence of PTAB activity is itself the signal. A patent this foundational in the molecular-barcoding space, held by a well-resourced assignee (Agilent Technologies, Inc.), and cited by later giants has never once been put in front of the Board. If you are being asserted against, you would be the first to challenge it at the PTAB — which means no estoppel against you, no prior petitioner's record to inherit, and full freedom to build the § 102/§ 103 record from scratch, but also no helpful FWD and only about four months of remaining term to work with.

Caveats, stated plainly: (i) this conclusion is drawn from the USPTO ODP block plus corroborating web/aggregator searches, not from an exhaustive PACER/PTAB-E2E/RPX/Unified Patents sweep — absence of a hit is strong but not conclusive proof that no proceeding exists; (ii) I did not retrieve the Certificate of Correction's content and am not representing what it says; (iii) the status of the October 2025 PTAB institution-policy changes and proposed rules as of 2026-09-28 is uncertain to me and should be verified before relying on them.

I'll attempt to retrieve the actual USPTO assignment records (reel/frame, correspondent) for this patent before writing the report.

Let me try to locate the actual reel/frame records and the corporate context for the PGT → Agilent transfer.

Let me make one more attempt to obtain reel/frame numbers and correspondent details directly.

Assignment & Ownership Chain — US 7,537,897 B2 ("Molecular Counting")

Coverage limitation stated up front (please read before relying on anything below): I was unable to retrieve the Assignment Center records themselves — i.e. the reel/frame numbers, execution dates, and correspondent-of-record attorney/firm for each link. The assignmentcenter.uspto.gov and assignment.uspto.gov/patent/index.html records are behind a session-based query interface that my searches did not surface in indexed form, and my final targeted attempt was cut off. I am therefore not fabricating reel/frame numbers. Every date, assignor and assignee below comes from (a) Google Patents' machine-readable legal-events/assignment index for US 7,537,897, and (b) Agilent's 24 July 2017 press release. Reel/frame and correspondent are marked NOT RETRIEVED wherever they would go. This materially weakens NPE signal 3 (repeat correspondent), which is the strongest detector in the framework and which I simply cannot run on this chain.

Verify at: https://assignmentcenter.uspto.gov/ or https://assignment.uspto.gov/patent/index.html — query "7537897" (and "11656830" for the pre-issue application record).

Inventors

Pattern note — and this one is a real finding. The relationship between the inventors and the assignee is inverted from the norm. On a typically-filed patent, the inventors assign to the operating company shortly after filing. Here, the inventors assigned their rights to COMPASS GENETICS, LLC — not to Population Genetics Technologies Ltd — on/around 2007-03-14, roughly seven weeks after the 2007-01-22 filing. They never appear as assignors to PGT at all. That is consistent with a founder/investor holding vehicle taking the founders' IP as the initial asset contribution, not with a routine employee invention-assignment. (Corroboration: "Compass Genetics Investors" is independently listed among Population Genetics Technologies' investors alongside Auriga, Wellcome Trust, Beringea and Syngenta Ventures.) Caveat: I am inferring the commercial purpose from the corporate-finance trail; I have not seen the Compass Genetics LLC operating agreement, so the "why" is inference, not evidence.

I see no evidence of the classic "all inventors departed within 12 months of filing" precursor to a portfolio fire-sale. Brenner remained publicly identified with PGT's portfolio at least until the 2017 sale.

Original assignee

Population Genetics Technologies Ltd. — Cambridge, United Kingdom (Salisbury House, Station Road, CB1 2LA per later filings).

• Status: UK private limited company No. 05116842, incorporated 30 April 2004, dissolved 5 January 2022 following voluntary liquidation (declaration of solvency filed 19 July 2019; full accounts filed March 2019). This was a solvent wind-down after the 2017 asset sale, not a bankruptcy. It is not an insolvency fire-sale of patents — the patents had already left the company four years earlier. Note the search-result ambiguity: one open-registry page dates incorporation to 30 April 2004 while Agilent's own release says the company was "founded in 2005." Both can be true; do not treat the discrepancy as significant.
• Primary line of business: never a volume product manufacturer. Formed to commercialise Sydney Brenner's NGS-detection IP. It licensed (VeriTag / molecular barcoding to Agilent, 2013) and by its own acquirer's description "in 2015, Population Genetics became an IP holding entity with over 30 patents." That is the company's own words in Agilent's press release.
• Did it ship a product embodying the claims? No product of its own that I can identify. The practising product is Agilent's — SureSelectXT HS, which Agilent states "incorporates molecular barcodes from this IP portfolio" and which was built on the pre-existing Agilent–PGT licence. So the claims are practised by Agilent, not by PGT.
• Acquired/dissolved: assets acquired by Agilent (announced 24 July 2017, financial terms undisclosed); the shell later liquidated in 2019–2022.

Assignment timeline

Google Patents' reassignment index shows four recorded transfers. Recorded dates are as indexed; execution dates and reel/frame numbers are NOT RETRIEVED and are omitted rather than guessed.

1. Recorded 2007-03-14 — Reel NOT RETRIEVED

• Conveyance: Assignment (indexed as "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
• Assignor: BRENNER, SYDNEY; MACEVICZ, STEPHEN C. (the inventors)
• Assignee: COMPASS GENETICS, LLC (as styled in the Google record — "COMPASS GENETICS, LLC"; note the mixed rendering elsewhere in the index)
• Correspondent: NOT RETRIEVED — no attorney, firm or address recoverable.
• Context: Founders'/investors' holding vehicle takes the inventors' rights as an initial asset contribution. Filed only ~7 weeks after the 2007-01-22 application date.

2. Recorded 2008-04-28 — Reel NOT RETRIEVED

• Conveyance: Assignment (internal reassignment; indexed "ASSIGNMENT OF ASSIGNORS INTEREST")
• Assignor: COMPASS GENETICS, LLC
• Assignee: POPULATION GENETICS TECHNOLOGIES LTD.
• Correspondent: NOT RETRIEVED.
• Context: Internal reorg — the holding vehicle returns/vests the rights in the operating parent ~13 months after taking them, ~13 months before issuance. Classic founder-IP consolidation, not a resale.

3. Recorded 2017-10-20 — Reel NOT RETRIEVED

• Conveyance: Assignment (indexed "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
• Assignor: POPULATION GENETICS TECHNOLOGIES LTD
• Assignee: AGILENT TECHNOLOGIES SINGAPORE HOLDINGS PTE. LTD.
• Correspondent: NOT RETRIEVED.
• Context: Acquisition / asset purchase. Matches Agilent's 24 July 2017 announcement that it "acquired the molecular and sample barcoding patent portfolios of Population Genetics Technologies," financial terms undisclosed, with ~30+ patents including this family. Recorded date is ~3 months after announcement — normal closing/recording lag.

4. Recorded 2017-10-20 (same day) — Reel NOT RETRIEVED

• Conveyance: Assignment (intra-corporate)
• Assignor: AGILENT TECHNOLOGIES SINGAPORE HOLDINGS PTE. LTD.
• Assignee: AGILENT TECHNOLOGIES, INC. (the NYSE-listed parent)
• Correspondent: NOT RETRIEVED.
• Context: Internal reorg — same-day upstream vesting of the acquired portfolio from the Singapore acquisition vehicle into the US public parent.

Cross-check against the previously generated sections: the ordering and dating of links 2–4 are consistent with the earlier "Patent summary" and "Litigation summary." One refinement worth flagging: the earlier sections used the shorthand "filed by Population Genetics Technologies Ltd → assigned to Compass Genetics, LLC," which reads as though PGT made the 2007 assignment. The index actually shows the inventors as assignors on that link, with PGT appearing only as assignee in 2008. PGT is listed as original assignee on the printed face of the patent; the assignment record tells the messier, two-step story.

Timeline diagram

timeline
    title Ownership of US 7537897
    2004 : Population Genetics Technologies formed
    2006 : Priority provisionals filed
    2007 : Application filed Jan 22
         : Inventors assign to Compass Genetics LLC
    2008 : Compass Genetics LLC assigns to Population Genetics Technologies
    2009 : Patent US 7537897 issues May 26
    2013 : Agilent license for molecular barcoding
    2015 : Population Genetics becomes IP holding entity
    2017 : Agilent acquires barcoding patent portfolio July 24
         : Recorded to Agilent Singapore Holdings Oct 20
         : Same day transfer up to Agilent Technologies Inc
    2019 : Population Genetics enters voluntary liquidation
    2022 : Population Genetics Technologies dissolved

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT (with one real caveat).
There is no operating-company-to-licensing-LLC hop anywhere in this chain. The exits are: founders → a founder/investor vehicle (2007-03-14), then up into the patent-holding parent (2008-04-28), then in 2017 into a publicly traded operating company and its Singapore acquisition subsidiary (both 2017-10-20). The names "COMPASS GENETICS, LLC" and "AGILENT TECHNOLOGIES SINGAPORE HOLDINGS PTE. LTD." look shell-like, but the concrete evidence points the other way: Compass Genetics Investors is documented as a PGT investor, and Agilent Singapore Holdings is a wholly-owned vehicle of an NYSE-listed company whose product line uses the technology. Caveat that is a genuine partial finding: PGT itself was an IP-holding entity by 2015 by its acquirer's own statement — but an IP-holder that licenses and then sells to a strategic acquirer is not a litigating shell, and there is no reeling/registered-agent indicia retrievable (no correspondent, no addresses retrieved). Marked not present, not "unclear," because the recipient of the patent is an operating company.

2. Known asserter in the chain — NOT PRESENT.
No assignee or assignor here (Population Genetics Technologies Ltd; Compass Genetics, LLC; Agilent Technologies Singapore Holdings Pte. Ltd.; Agilent Technologies, Inc.) matches Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. I also did not find any Unified Patents or RPX high-frequency-plaintiff listing for any of the four names. Limit: I did not query RPX Insight or the Unified Patents portal directly.

3. Repeat correspondent across the chain — UNRETRIEVED / CANNOT ASSESS.
This is the signal the assignment records are most valuable for, and it is exactly the field I could not obtain: no correspondent name, firm or address was retrievable for any of the four recordings. I will not name a firm by inference. This is an open gap, not a negative finding.

4. Cascading transfers — NOT PRESENT as an NPE pattern; present but benign in form.
Two hops occur within ~13 months (2007-03-14 → 2008-04-28), both pre-issuance, through a founder/investor vehicle that then vesting rights in the operating parent. The 2017 pair occur on the same recorded date and are an intra-corporate upstreaming from an acquisition subsidiary to its listed parent — not a chain of unrelated LLCs. No shared-correspondent or common-principal evidence is retrievable, so I cannot positively exclude a common intermediary, but nothing observed resembles serial flipping.

5. Pre-litigation transfer — NOT PRESENT. No infringement suit naming 7,537,897 as asserted patent was found in the previously generated litigation section, so there is no pre-suit assignment window to test. The 2017-10-20 recordings postdate no identified complaint. Caveat carried over: absence of a found suit is a search-coverage limit, not proof of non-assertion.

6. Bankruptcy fire-sale — NOT PRESENT. Population Genetics Technologies Limited entered voluntary liquidation with a declaration of solvency (19 July 2019) and dissolved 5 January 2022. That is a solvent members' voluntary liquidation, the opposite of a Chapter 7/11 distress sale — and, critically, the patents were sold in 2017, ~2 years before liquidation began, so this patent was not liquidated inventory. No Kodak/Nortel/Polaroid-style event in this chain.

7. Privateering — NOT PRESENT. Privateering requires the operating company to park IP with an NPE that then asserts for it. Here the transfer ran in the opposite direction: the patent left a non-practising IP holder and landed at the operating company that sells the practising product (SureSelectXT HS). Agilent is the buyer/practitioner, not a backer of an assertion vehicle.

8. Defensive aggregator — NOT PRESENT (technically), but the functional outcome is similar. The chain does not terminate at RPX, AST, LOT Network, Unified Patents or OIN — it terminates at Agilent Technologies, Inc. The patent is therefore not formally neutralised; a listed operating company could in principle assert it. There is simply no record of it being asserted.

Verdict

Defensive / non-asserting — closest available fit, with the definitional caveat stated.

Agilent is not a defensive aggregator, so the verdict label does not match its own definition precisely. It is nonetheless the best-supported choice among the five, and this is why: the chain runs founders → founder/investor vehicle → patent-holding parent → NYSE-listed operating company (Agilent Technologies, Inc., per the two 2017-10-20 recordings), with zero NPE signals present, a practising product in Agilent's SureSelectXT HS line, and no assertion of this patent found in the prior litigation search. "Operating-company assertion" is not available because that verdict requires the current assignee to be suing actual competitors, and no such suit exists on the record. "Insufficient data" is not available either, because the chain itself is well documented beyond the original assignment.

Two things would change this call, and both are open:

1. The correspondent data. If the Assignment Center correspondent for the 2017-10-20 recordings or the 2007-03-14 recording turns out to be a known NPE-recording attorney, signal 3 fires and the picture shifts. I could not run this test.
2. Undiscovered litigation. If a 2026 assertion naming 7,537,897 exists but is not indexed by the aggregators searched, signal 5 becomes live and the verdict moves toward privateering-type analysis (Agilent asserting through a subsidiary against a competitor would be operating-company assertion, not NPE — but a transfer to a third-party vehicle would be different).

Recommended verification: pull the four reel/frame records directly at https://assignmentcenter.uspto.gov/ for patent 7537897 and application 11/656,830, capture the correspondent of record on each, then re-run signal 3 by cross-checking those attorney names against Unified Patents and RPX asserter lists. I have deliberately left those fields blank rather than filling them.

I'll search for the USPTO record and the cited references for this specific patent.

I found the front-page "References Cited" section. Let me verify and expand the list with additional targeted searches.

I have the front-page reference list. Let me verify the titles and dates of the key cited U.S. patents.

Prior-Art Analysis — US 7,537,897 ("Molecular counting")

A. Scope note and a critical methodological caveat

I searched for the USPTO/Google Patents record and for the printed front page ("References Cited") of US 7,537,897. I recovered the front-page citation list from two independent reproductions of the printed patent (the Docket Alarm PTAB exhibit of the patent, and the patent copy filed as 10x Genomics v. Celsee, D. Del. No. 1:19-cv-00862, Doc. 104‑3). Both agree on the reference numbers, which raises confidence in the identifiers.

Two caveats you must weigh:

1. A U.S. patent front page does not contain a per-claim relevance table. The "relevant to claim No." column you are asking for exists in EPO/PCT search reports, not on a U.S. patent face. Therefore the examiner's own claim-by-claim mapping for US 7,537,897 is not retrievable from the patent itself. Everything in Section D below is my analytical inference under 35 U.S.C. § 102, not a quotation of the examiner's record.
2. The source text is OCR of a scanned front page, so some dates/names are garbled (I flag these). I hit my tool budget before individually verifying every title, so titles I have not confirmed in this session are explicitly marked [unverified].

I also note (consistent with the two previously generated sections) that no litigation or PTAB challenge naming '897 as the patent-in-suit was found; the only contested-proceeding appearance is as a prior-art exhibit (Exhibit 1058) in IPR2023-00955.

B. The patent record (from the authoritative text; not repeated here)

US 7,537,897 B2, "Molecular Counting," Brenner & Macevicz, Appl. 11/656,830, filed 2007‑01‑22, priority 2006‑01‑23, granted 2009‑05‑26; 20 claims; five independent claims (1, 8, 12, 17, 20). Because the application was filed 2007‑01‑22, every printed publication dated on or before ~2006‑01‑22 is available as pre‑AIA § 102(b) art.

C. The cited "U.S. Patent Documents" (front page)

All of the following were located on the '897 front page. Dates/numbers are as printed; [unverified] = title not independently confirmed in this session.

⚠ Discrepancy to flag: the front-page OCR shows 6,440,667 (Fodor), whereas the specification of the same patent expressly cites "Lipschutz et al., U.S. Pat. No. 6,440,677" for microarray synthesis. These are two different numbers. Per the instruction to interpret identifiers literally, I have not auto-corrected either; the correct identity/date of the front-page entry should be confirmed directly.

D. The cited "OTHER PUBLICATIONS" (front page)

Also cited within the specification (not necessarily on the front page): Brenner PCT WO 2005/080604 (sorting-by-sequence); Macevicz PCT WO 2005/111242 (padlock probes); Margulies et al., Nature 437:376‑380 (2005) (emulsion PCR/pyrosequencing). These matter for the dependent claims.

E. § 102 analysis by independent claim

Timing. Every U.S. patent above issued on or before April 2003, i.e. more than one year before the 2007‑01‑22 filing → all are available as § 102(b) art. So are the 1992/1998/1999/2000/2004 non‑patent publications. Daser et al. 2006 is the exception: published after the 2006‑01‑23 priority, it is not § 102(b) art, and it is § 102(a) art only against claims whose subject matter is not supported by the earlier provisionals (i.e., claims resting solely on the later provisionals such as 60/824,456 of 2006‑09‑04). Treat it as a § 102(a)/§ 103 reference against later-supported dependent claims, not against claim 1.

Claim 1 (tag = concatenation of subunits from a 2–6 member set; divide into aliquots by sorting on subunit position; repeat until an aliquot is empty).

• Best candidates: the Brenner oligonucleotide-tag/sorting family — 5,604,097; 5,635,400; 5,654,413; 5,714,330; 5,763,175; 5,846,719; 6,138,077; 6,140,489; 6,172,214; 6,235,475. These teach tags built from ordered "word" subunits and separating subsets by sorting on subunit identity — i.e., the structural and sorting elements.
• Encoded combinatorial chemistry (1992) independently supports the "tag = concatenation of words from a small defined set" element.
• Anticipation assessment: I could not confirm in this session that any single one of these discloses the full claimed counting endpoint ("repeat until at least one aliquot has no tags"). If a single reference does, claim 1 is anticipated; otherwise the likely posture is § 103 over a Brenner tag/sorting patent in view of a counting reference (e.g., Vogelstein 6,440,706).

Claim 8 (sorting tag + identification tag; e.g., binary tags + metric tags).

• 5,846,719 (Brenner) — expressly cited in the specification for "labeling by sampling"; the strongest § 102 candidate for the "labelled-by-sampling with a unique tag" element.
• Affymetrix/Chee/Fodor array patents (5,424,186; 5,445,934; 5,744,305; 6,355,431; 6,355,432) and the MPSS 2000 paper supply the "identify tags by array/sequencing readout" element.
• Anticipation assessment: the two‑tag architecture (a sorting tag plus a separate identification tag) does not appear to be disclosed by any single cited reference → probable § 103, not § 102.

Claim 12 (a plurality of differently tagged probes per target; nuclease‑resistant selected probe; isolate with nuclease; sequence the tags and count).

• 5,854,033 (Lizardi) and Lizardi et al. 1998 — circular/rolling-circle probes and single-molecule counting.
• 6,117,631 (Nilsson) — padlock/probe circularization chemistry.
• Callow et al. 2004 — adapters conferring nuclease-resistant selected products; the specification itself relies on Callow for this.
• Anticipation assessment: these references cover the probe/circularization/nuclease mechanics well; none appears to disclose the specific combination of (i) a probe plurality per target, (ii) one unique tag per selected probe, and (iii) tag sequencing to count. Likely § 103.

Claim 17 (label by sampling → amplify tags → sequence to count).

• MPSS 2000 (Brenner) discloses amplifying and sequencing tags for digital measurement; 6,440,706 (Vogelstein) discloses digital counting of single molecules.
• Anticipation assessment: this is the claim most exposed to a § 102/§ 103 challenge, since "amplify tags and sequence them to count" is squarely the MPSS/digital-counting theme.

Claim 20 (label with metric/length-encoded tags → amplify → separate into bands → count bands).

• Anticipation assessment: I found no cited reference disclosing counting distinct electrophoretic bands of length-encoded tags as the counting readout. The metric-tag concept appears to be applicant-originated. Low § 102 exposure among the listed art.

Dependent claims.

• 2–7 (binary/dinucleotide tags) → 1992 PNAS and the Brenner tag patents.
• 9–11 (binary tag = dinucleotide concatenate; metric identification tags) → same, plus the applicant's own metric-tag work.
• 13–16 (restriction-fragment targets, circularizing adaptors, type‑IIs/"double cleavage" type‑IIs, concatenation) → Callow 2004 and the Lizardi/Nilsson circularization art.
• 18–19 (emulsion PCR; pyrosequencing/Sanger/ligation sequencing) → Margulies et al. 2005, Nature 437:376‑380 (cited in the specification).

F. Bottom line

• The front page of US 7,537,897 cites ~37 U.S. patents (1988–2003), all § 102(b)-eligible, and ~7 non-patent publications, dominated by Brenner's own prior work (MPSS, encoded combinatorial chemistry, the oligonucleotide-tag/sorting patent family) plus Lizardi (RCA/padlock), Callow (double-sided adapters), Vogelstein (digital amplification), and Daser (MCC).
• The most threatening references are the Brenner tag/sorting patents (esp. 5,846,719) and Brenner et al. 2000 MPSS, for the tag-structure, sorting, and sequence-counting elements; Lizardi 5,854,033 / Lizardi 1998 and Callow 2004 for the probe/nuclease elements; Vogelstein 6,440,706 for the digital-counting concept.
• No single front-page reference appears to disclose every element of any independent claim, especially the two‑tag (sorting + metric) architecture of claim 8 and the length‑band metric-tag counting of claim 20 — so the realistic posture is § 103 combinations, with claim 17 the most § 102-exposed.
• Daser et al. 2006 is the one listed item that post-dates the priority date and should be treated as § 102(a)/§ 103 only against later-supported claims.

What I could not do (stated plainly): I did not verify, in this session, the individual titles/dates of several references (marked [unverified] above), did not confirm whether any single Brenner patent recites the "sorting until an aliquot is empty" endpoint, and did not resolve the 6,440,667 vs. 6,440,677 discrepancy. To firm this up, the front page should be read directly against Google Patents → US 7,537,897B2 → "Citations" and against the USPTO PatentCenter "References Cited" / File Wrapper for Appl. 11/656,830, which together will give the authoritative list (and any foreign patent documents not captured by the OCR I retrieved).

I'll ground the § 103 analysis in the actual prior-art references cited/incorporated by this patent and in the contemporaneous literature. Let me pull primary sources.

Good primary material. Let me pull the key "labeling by sampling" and digital-karyotyping/counting references, plus the patent's own cited-references list.

Obviousness Analysis — U.S. Patent No. 7,537,897 ("Molecular Counting")

Statute applied: Pre‑AIA 35 U.S.C. § 103 (application filed 2007‑01‑22, before the 2013‑03‑16 AIA change), with Graham v. John Deere factors as refined by KSR Int'l v. Teleflex (2007). Critical date: the § 102(a)/(b) reference date is the priority date of 2006‑01‑23 (earliest provisional 60/761,578) — not the 2007 filing date.

Note on the "Prior Art section." The Google Patents page as fetched shows a partial OCR of the face‑of‑patent References Cited list and the Cited By list, plus the specification's own incorporated‑by‑reference documents. The OCR was truncated (I could read U.S. patents through roughly Winston 6,544,739, but not the complete "Other Publications" list). I therefore treat as the prior‑art corpus: (i) the front‑page U.S. patent citations actually legible; (ii) the references the specification expressly incorporates by reference; and (iii) contemporaneous literature pre‑dating 2006‑01‑23. I could not retrieve a verbatim claim set from the authoritative dump (it truncates before the claims), so claim limitations below are taken from the previously generated claim summary, which I treat as authoritative per instructions. This is a genuine evidentiary gap and I flag it rather than paper over it.

1. Person of ordinary skill in the art (POSA)

A POSA here is a molecular biologist/genomicist (Ph.D. or M.D./Ph.D. with ~2–3 years' experience, or equivalent) familiar with: oligonucleotide tagging, the Brenner "minimally cross‑hybridizing set" and "labeling‑by‑sampling" literature; type IIs restriction enzymes; ligation/padlock probes; exonuclease purification; PCR/emulsion PCR; and the standard sequencing and electrophoretic readouts of the day. That POSA would have been the same person who read Brenner '719, the SAGE papers, and digital karyotyping as a matter of routine.

2. Prior‑art inventory (grounded citations)

Two of these are not merely found by me — the patent itself admits them:

• "Preferably, unique tags are attached to the molecules to be counted by a process of labeling by sampling, as described by Brenner et al, U.S. Pat. No. 5,846,719." (spec.)
• "…tag sequences can be sorted … using the sorting‑by‑sequence technique disclosed in Brenner, PCT publication WO 2005/080604, which is incorporated herein by reference." (spec.)

Specification admissions of this kind are usable as evidence of what the art already contained. (Caveat: incorporation by reference does not itself make a document prior art unless it was publicly available before the critical date — but A and B both were.)

3. Claim‑by‑claim § 103 analysis

Ground I — Claims 1, 17 (and 8) over A + B + C/D + G

• Limitation (a) — "each different molecule attached to a different tag; tag = concatenation of subunits from a set of 2–6 members": fully disclosed by A (labeling by sampling; subunit concatenations from a defined/minimally cross‑hybridizing set).
• Limitations (b)/(c) — divide into aliquots by sorting on the subunit at the first/successive position; repeat until an aliquot is empty: disclosed by B (successive sequence‑specific partitioning that monotonically reduces complexity). The patent's own text supplies the algorithm ("number of molecules proportional to 2^r"), i.e., the very counting rule — an admission that the sorting technique is A/B, and the only added step is to read the sorting count as a molecule count.
• Claim 17 — amplify tags, then determine the number of different tags by sequencing: C and D expressly do exactly this (tally tags after amplification/concatemerization by sequencing) in the express contexts the '897 patent names as its applications ("genetic copy number variation, aneuploidy, … gene expression changes"). G supplies clonal/bead sequencing and pyrosequencing for the dependent claims (emulsion PCR, pyrosequencing).
• Claim 8 — sorting tag + identification tag, count = (sorting rounds) + (different tags in the sorted mixture): This is a two‑tag architecture, i.e. A/B's sorting tags (B) combined with a readout tag. Using one tag domain for the partitioning and a separate, simpler domain for the final readout is the predictable substitution of a readout element (KSR rationale (b)); A already teaches tag populations of different complexity used for different functions, and H/C/D teach scalar/countable readouts.

Ground II — Claim 12 (nuclease‑resistant "selected probe") over E + F + A + G

• (a) "for each target polynucleotide a plurality of probes, each with a different tag": E discloses target‑specific ligation/padlock probes (with tags); A teaches making the tag repertoire's complexity far exceed the target number so each captured target gets a unique tag. Providing a plurality of differently tagged probe molecules per target is the direct, predictable implementation of A's "repertoire ≫ targets."
• (b)/(c) "associate → selected probe resistant to a nuclease → isolate by nuclease treatment": E (circularization/ligation only on target) + F (exonuclease removal of non‑circularized species) — the exact combination is routine and was already used together in the padlock/rolling‑circle art.
• (d) sequence tags in a sample of isolated probes, count: C/D/G.
• Dependent claims (restriction‑fragment targets with unique overhangs; circularizing adaptor forming a dsDNA circle; type IIs and "double cleavage" type IIs such as Bae I; concatenation before sequencing): the type‑IIS/overhang machinery is standard (Szybalski et al., Gene 100:13‑26 (1991), cited in the spec) and concatenation is C-SAGE.

Ground III — Claim 20 (metric tag → separate → count bands) over A + H + I

• "label by sampling each target with a unique metric tag": A.
• "amplify; separate to a profile of distinct bands; count the bands": Sizing of differently‑length PCR products on gels/PAGE and counting bands is among the oldest operations in molecular biology; H (Church 1988; Ohlmeyer 1993 binary positional encoding read out by a scalar property) supplies the "encode identity in a measurable physical property other than sequence" concept; C supplies "count the separated species to quantify molecules."
• Weakest link — flagging honestly: I did not locate a pre‑2006 reference that uses the term "metric tag" (length‑encoded identification tag paired with a binary sorting tag). The term appears to be the applicants' coinage (Table I of the '897 patent converts binary tags to metric tags). Claim 20's obviousness therefore rests on combining "known length readout + known counting" rather than on a single squarely‑on‑point reference; an examiner would need to articulate the motivation carefully, and the "metric tag" label could be argued as a non‑obvious nomenclature choice — although labels do not confer patentability.

Dependent‑claim clusters

• Binary tag = dinucleotide concatenate; no two subunits the same; set = {AG, AC, TG, TC}; non‑self‑complementary subunits → routine design choice / optimization of a result‑effective variable (In re Boesch; In re Aller). The "comma‑less"/minimally‑cross‑hybridizing rationale is squarely from A.
• Ranges (tag repertoire ≥10× or ≥100× target number; 90/95/99% uniqueness probability; 14–17 subunits for 10⁴–10⁵ tags) → arithmetic consequences of A's "complexity ≫ targets" teaching; obvious optimization.
• Concatenation/cloning (SAGE, C); emulsion PCR + pyrosequencing (Dressman 2003; Nyren/Ronaghi; Margulies 2005) → G.

4. Motivation to combine (KSR rationales)

1. Same field, same problem, overlapping inventorship. A, B and the '897 patent share Brenner/Macevicz and the same assignee family (Compass Genetics / PGT). A POSA would not treat them as disparate art; they are a continuous research program, and the '897 specification itself chains them together.
2. Art‑recognized problem with a known, finite set of solutions. Counting molecules in limited samples (single cells, fetal DNA, aneuploidy, methylation, expression) was a recognized need (spec.'s own framing). The art offered a small menu of readouts — hybridization array, gel sizing, Sanger, pyrosequencing — and KSR makes selection among a finite number of predictable options obvious.
3. Predictable combination of known elements. "unique labeling by sampling" + "probe/ligation capture" + "nuclease purification" + "amplification" + "sequence or size readout" are each known; combining them yields nothing more than the sum of their expected functions (KSR rationales (a), (b), (c)).
4. Conversion of a counting problem into a tag‑counting problem is taught by C and D. SAGE and digital karyotyping already establish that counting tags ≙ counting molecules; extending that to "count the molecules directly" is the natural next step, not an inventive leap.
5. Reasonable expectation of success. The '897 specification's own worked Example I (Table II lengths; FIG. 8C gel) demonstrates that recoding binary tags into length‑distinguishable metric tags "worked" — evidence that the combination was predictable in the art, not fraught with unpredictability.
6. Ullman/Gullberg (J) show the same tag‑and‑count architecture transferred to non‑nucleic‑acid targets (antibodies), supplying motivation for the "any molecule that can be tagged" breadth of claims 1/12.

5. Anticipated rebuttals and secondary considerations

• "The claims are a new use of known tags (counting rather than identifying)." KSR: a new, predictable use of a known composition/technique, where the art already suggested tag counts as abundance measures (C, D), is obvious. The sorting‑count algorithm ("2^r") is arithmetic on a disclosed process.
• "No single reference teaches it all." Under KSR, that is not required — motivation may come from the problem, the prior art as a whole, or common sense; and the specification's own admissions make A+B a de facto single teaching.
• Teaching away? None found. A's caution that molecules sharing tags cause undercounting is a design caution, not a teaching away; it instructs the POSA to size the repertoire — which the '897 dependent claims do.
• Secondary considerations / nexus. To date I found no litigation in which '897 is asserted (see prior sections), so there is no adjudicated record of long‑felt need, unexpected results, or copying. Any such argument would also face a nexus problem, since the alleged commercial value lies in the application (copy‑number/aneuploidy analysis) already practiced and disclosed by D and I.
• Potential "own work" issue. WO 2005/080604 (B/published 2005‑09‑01) and the Macevicz/Brenner applications US 7,217,522 (filed 2005‑02‑10) and US 7,393,665 (filed 2005‑07‑07; pub. US 2006/0177833 A1) predate the critical date. A § 103(c) common‑ownership argument (Compass Genetics/PGT) could only disqualify those that qualify solely under § 102(e)/(f)/(g); it does not reach § 102(a)/(b) patents/publications. So A and B remain available irrespective of common ownership, and the § 103(c) question is confined to the § 102(e)‑type applications.

Bottom line: Independent claims 1, 8, 12 and 17 are, on the record available, strongly vulnerable under § 103 over Brenner '719 + Brenner WO 2005/080604, reinforced by SAGE/digital karyotyping for the "count tags by sequencing" limitation and by padlock‑probe/exonuclease art for claim 12. Claim 20 is vulnerable but on a weaker combination (no pre‑2006 "metric tag" reference located); its defense would likely reduce to arguing the coined nomenclature and the specific tag‑design parameters, which are the kind of routine optimization KSR and In re Boesch discount.

6. Confidence and limitations (stated explicitly)

• High confidence: identity of A (US 5,846,719 — verified via PubChem and US 6,235,475), B (WO 2005/080604 A3, published 2005‑09‑01 — verified via the WIPO PDF), C (SAGE, Science 270:484 — verified via PubMed), and the fact that the '897 specification expressly names A and B as the source of its labeling‑by‑sampling and sorting‑by‑sequence steps.
• Moderate confidence: the precise § 102 sub‑category of each reference (a/b/e) and the § 103(c) analysis; and the mapping of B's disclosure to every element of claim 1 step (b)/(c), since I read the B abstract/summary rather than the full document line‑by‑line.
• Low confidence / flagged: (i) verbatim claim language (not in the supplied authoritative text; I relied on the prior section's claim summary); (ii) the complete "Other Publications" citation list on the '897 face (OCR truncated); (iii) any single reference disclosing a "metric tag" as such.
• Cross‑reference note (contradiction check): nothing in this analysis contradicts the earlier sections. One addition: a USPTO PTAB petition document (petitions/1558283) cites a reference numbered EX1014 whose quoted text — "a set of 64 double stranded composite tags… each composite tag contains a double stranded metric tag" (at 28:37‑39), plus "substantially every molecule to be counted in a sample…is associated with a probe having a different tag" (5:17‑21) and "a size of tag population is selected that ensures…a unique tag" (18:2‑4) — appears to be US 7,537,897 itself, cited as prior art against a different ("Lo") patent. This is a second instance (distinct from Exhibit 1058 in IPR2023‑00955) in which '897 is used as evidence rather than asserted. I flag it as "appears to be" because I did not open EX1014 to confirm its cover page.