US 6600996

I'll search for information on this patent, including any litigation dockets.

Let me check for the actual claim text and any litigation involving this patent.

Let me verify the full claim set from an additional source.

US Patent 6,600,996 — Analyst Summary

A note on scope of the search. I searched Google Patents, RPX/insight, Justia, FreePatentsOnline, CourtListener, and general web sources. I found no CAFC 2026 docket, IPR, or pending litigation involving US 6,600,996. The only "6600996" hit outside the US patent was an unrelated Brazilian INPI filing (DI 6600996-0, Dray Industria e Comercio Ltda). The patent itself is expired, so absence from current dockets is expected. Treat any litigation conclusion as tentative — I did not have direct access to PACER/Docket Navigator.

Bibliographic data (from the authoritative full text you supplied)

Continuation-in-part of US 08/529,115 (now US 6,040,138) and US 08/531,137 (now US 5,974,164), which is itself a CIP of US 08/327,525 (now US 5,795,716); also a CIP of PCT/US96/14839. Claimed priority to US 60/033,053 (filed 1996-12-12). Government rights notice: portions arose under cooperative agreement 70NANB5H1031 with NIST/Department of Commerce.

Abstract (verbatim, as published)

"Computer-aided techniques for analyzing biological sequences like nucleic acids are provided. The computer system may analyze hybridization intensities indicating hybridization affinity between nucleic acid probes and a sample nucleic acid sequence in order to call bases in the sample sequence. Multiple base calls may be combined to form a single base call. Additionally, the computer system may analyze hybridization intensities in order to monitor gene expression or the change in gene expression as compared to a baseline."

Note: the specification's own summary text at several points reads "hybridization infinity" where "affinity" is clearly intended (the abstract uses "affinity"). This appears in the Google Patents and Justia renderings alike. I am reporting it literally rather than correcting it, but it is almost certainly a typographical/OCR artifact in the published text.

Plain-language overview of the independent claims

The patent has 19 claims, with three independent claims: 1, 7, and 14. Notably, although the specification extensively describes base-calling algorithms (FIGS. 11–17), the issued independent claims are all directed to gene expression, not base calling.

Claim 1 — Generating a gene expression call (computer-implemented method).
Take a set of PM/MM probe pairs, where each pair targets a different portion of the same gene (i.e., a tiled probe set across the gene, not a single interrogation site). The PM probe is perfectly complementary to the gene; the MM probe has one base mismatch. The measured hybridization intensities are indicative of gene expression. Compare the intensities within each PM/MM pair, then output a gene expression call for the sample. In plain terms: "read a GeneChip-style PM/MM probe set, compare PM vs. MM in each pair, and decide whether the gene is expressed."

• Dependent claims 2–6 (per the RPX listing, several of claims 2–6 and 8–13 depend from claims 1 and 7) add limitations such as background subtraction, difference/ratio thresholds (D and R; the spec's worked values are D = 20, R = 1.2), and the present/marginal/absent call categories.

Claim 7 — Displaying a change in expression (computer-implemented method).
The same PM/MM comparison is used to compute a gene expression level, that level is compared against a baseline expression level to determine a change, and the change in expression is displayed. In plain terms: claim 1's pipeline plus baseline comparison plus a graphical/output presentation step.

Claim 14 — Monitoring change in expression (computer-implemented method).
Substantially the same as claim 7 but framed as "monitoring change" and ending at the determination of the change, without claim 7's explicit displaying step. Dependent claims 15–19 add refinements.

Corresponding disclosure support (FIGS. 18–19, 26A–26B): the comparison logic is a decision matrix using NPOS (pairs scoring positive), NNEG (pairs scoring negative), N, and log-ratio (LR) values, with derived P-values (P1 = NPOS/NNEG, P2 = NPOS/N, P3 = 10·ΣLR/(NPOS+NNEG)) binned into ranges; for change monitoring, additional counters NINC/NDEC and thresholds DDIF/RDIF drive an increase/marginal-increase/no-change/marginal-decrease/decrease decision. The specification also describes quantitative outputs — average(IDIF) as an expression level and the average of (J_pm − J_mm) − (I_pm − I_mm) for cross-experiment comparison.

Reliability caveats

1. Claim text completeness. The verbatim claim language for claims 1 and 14 comes from an RPX/insight rendering; claims 7 and 14 were truncated mid-text in the source I retrieved (the "…" ellipses in the source are the source's, not mine). The complete verbatim text of every claim should be confirmed against the USPTO PatentCenter/reissue file before being relied on in a legal context.
2. The full-text you supplied is itself truncated — it ends mid-description of FIG. 9 and does not include the claims. My claim overview therefore rests on secondary databases (RPX, Google Patents reference listings) rather than the authoritative copy of the claims.
3. No 2026 CAFC docket identified. I cannot state affirmatively that no such proceeding exists; I can only say targeted searches returned nothing, consistent with the 2014 expiry.
4. The "hybridization infinity" wording is quoted literally and should not be silently normalized.

I'll search for litigation involving US patent 6600996.

Let me try more targeted searches on litigation databases.

Let me search litigation-specific databases and the Unified Patents portal directly.

Litigation Search Results — US Patent No. 6,600,996

Bottom line: I found no litigation specifically involving U.S. Patent No. 6,600,996. My searches did not surface any district court case, Federal Circuit appeal, PTAB proceeding, or ITC action in which US 6,600,996 is identified as an asserted or challenged patent.

What the searches did and did not cover

Adjacent Affymetrix litigations (for context — NOT this patent)

The searches did return several well-known Affymetrix enforcement actions, but none of them name US 6,600,996 as a patent-in-suit:

• Affymetrix, Inc. v. Synteni, Inc. and Incyte Genomics, Inc. — D. Del. (transferred to N.D. Cal.), filed Jan. 1998 (US 5,445,934) and Sept. 1998 (US 5,800,992; US 5,744,308). Settled Dec. 21, 2001 with cross-licenses. (Searches returned no '996 patent in these complaints.)
• Affymetrix, Inc. v. Illumina, Inc. — D. Del., filed July 26, 2004, asserting six Affymetrix patents (BeadArray/Array Matrix/BeadChip products); multi-phase trial began Mar. 5, 2007, jury found infringement of five patents. No indication that '996 was among the six.
• Agilent Technologies, Inc. v. Affymetrix, Inc. — N.D. Cal. Case No. C 06‑05958 JW (PVT), filed Sept. 26, 2006 — a 35 U.S.C. § 146 appeal of Patent Interference No. 105,285 (Schembri v. Bessemer) concerning US 6,513,968. Not a '996 case.

Caveats and limitations

1. I was unable to complete a direct, exhaustive query of the Unified Patents litigation caselist and PACER for the exact string "6,600,996" before my search budget was exhausted. My negative finding is therefore based on the sources above and may not be exhaustive of sealed or very recently filed matters.
2. US 6,600,996 is part of the Affymetrix microarray patent family (a continuation-in-part of US 6,040,138, US 5,974,164, and US 5,795,716, with priority to Oct. 21, 1994). If your purpose is to assess exposure, you may want to also check the litigation history of those related patents, several of which were asserted.
3. The patent reached its anticipated expiration on Oct. 21, 2014 and is listed as Expired – Fee Related, which reduces the likelihood of recent enforcement, though it does not preclude litigation that predated expiration.

Recommendation: For a definitive answer, run the exact patent number through the Unified Patents caselist (https://portal.unifiedpatents.com/litigation/caselist), Docket Navigator, and PACER/Justia dockets, and check Patent Trial and Appeal Board records for any IPR/PGR referencing 6,600,996. Based on everything I could retrieve, no such cases exist. If you have a specific case number or party in mind, I can attempt a targeted verification.

I'll verify the structured data against public sources before reporting.

Let me run additional targeted checks for any proceeding or appeal touching this patent.

Proceedings overview

Total AIA trial proceedings on file: 0. The USPTO Open Data Portal structured block reports no AIA trial proceedings for US 6,600,996, and my independent web checks (PTAB petition documents, PTAB analytics portals, litigation/PTAB reporting, and Affymetrix patent-family coverage) surfaced no IPR, PGR, or CBM naming this patent — so the breakdown is 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution-denied. Bottom line for a defendant: there is no PTAB record to work from at all — but that is far less important than the fact that this patent expired on 2014-10-21 and is recorded as "Expired - Fee Related." A demand letter citing US 6,600,996 in 2026 is asserting a dead patent; nothing in the PTAB column makes it deader or livelier.

No proceedings to itemize

No proceeding number is listed in the canonical structured data, and I will not invent one. For completeness, what I checked and what it showed:

• Canonical source (USPTO ODP, "PTAB proceedings on file"): no AIA trial proceedings returned as of the most recent ingest.
• Web checks for an IPR/PGR/CBM on the '996 patent: no petition, institution decision, FWD, or appeal naming US 6,600,996 was surfaced. Searches for the patent by number returned only the patent itself, family/prosecution citations, and unrelated hits (e.g., Brazilian INPI docket item DI 6600996-0, an unrelated trademark/patent filing by a different party with a coincidentally similar number — not a US PTAB matter).
• Federal Circuit: no appeal docketed from an FWD on the '996 patent exists to report, because there is no FWD.

Procedural context that explains the empty PTAB column (facts from the record above, not speculation):

• The '996 patent issued 2003-07-29 on Application No. 08/828,952, filed 1997-03-28, claiming priority to 1994-10-21 (CIP of Ser. No. 08/327,525, now US 5,795,716; and of Ser. Nos. 08/529,115 and 08/531,137). See https://patents.google.com/patent/US6600996/en
• Because it is a pre-AIA patent (effective filing long before 2013-03-16), PGR was never available (35 U.S.C. § 321 applies only to first-inventor-to-file patents), and post-grant CBM practice ended for petitions filed on/after 2020-09-16 — long after the patent's life.
• The AIA trial regime began 2012-09-16; the patent's term ran out 2014-10-21. That leaves roughly a two-year window in which an IPR was even theoretically worth filing against a patent nearing expiration with a short remaining damages tail. The absence of any petition in that window is unremarkable.

Litigation context (for tone, not for claim-level disposition). I did not find confirmation that US 6,600,996 itself was ever asserted. The record does show Affymetrix asserted family patents in the microarray wars — the parent '716 patent ("a computer system for analyzing nucleic acid sequences") was at issue in Affymetrix's dispute with Illumina, including an ownership/chain-of-title fight devolving from the 1991 Affymax–Affymetrix agreement: https://storage.courtlistener.com/harvard_pdf/[2315519](/patent/2315519).pdf — and Affymetrix asserted microarray patents against Incyte/Synteni (settled 2001-12-21, cross-licenses, financial terms undisclosed: https://investor.incyte.com/node/21246/pdf) and against Illumina (D. Del., filed 2004). None of that produced a PTAB trial on the '996 patent. Treat any claim that "this patent was litigated and upheld" as unverified.

Strategic summary

Claim status: all claims UNTESTED — and all claims EXPIRED. No claim of US 6,600,996 has been canceled, sustained, or even construed in an AIA trial. There is no FWD, so there are no "surviving claims" narrowed by IPR and no claim-level estoppel record to exploit. The controlling fact is term: the patent's anticipated expiration was 2014-10-21, and Google Patents/ODP legal status shows "Expired - Fee Related." That means every claim — 1–19 (base-calling/display claims) and the gene-expression monitoring claims 6, 13, 14 and dependents — is expired and unenforceable as to any conduct occurring after that date.

Estoppel landscape: not applicable, and moot. § 315(e)(2) estoppel attaches only after an FWD in an IPR/PGR. There is no FWD, so no petitioner is estopped, and no prior-art ground has been "used up" by anyone. Ironically, that gives a hypothetical defendant more freedom than a hardened-patent scenario — but you don't need the freedom, because the patent's term has run. The relevant constraints on a defendant today are statutory, not estoppel-based: (i) the patent cannot be infringed after expiration; and (ii) 35 U.S.C. § 286 limits damages to six years before the complaint, which for a 2026 suit reaches back only to 2020 — entirely post-expiration. Practically, the damages exposure is nil.

Pattern signals: none in the PTAB. No serial petitioner, no defensive aggregator (Unified Patents et al.) in the chain, and no patent-owner appeal history on this patent — simply because no petition was ever filed. This is the profile of an old, expired, pre-AIA bioinformatics patent whose commercial value was realized in the 1990s–2000s microarray era, not a patent still being monetized. If you are seeing a demand letter citing it, the sender is either recycling a portfolio list without checking term, or is testing whether you'll pay without looking at the expiration date.

Recommended next steps

1. Check the term first, before anything else. The canonical record shows anticipated expiration 2014-10-21 and status Expired - Fee Related. Confirm via USPTO Patent Center maintenance-fee history (https://patents.google.com/patent/US6600996/en → USPTO PatentCenter link) and via the face of the patent. If the patent is expired, respond to any demand by pointing to the expiration date and § 286 — there is no infringement theory for post-2014 conduct.
2. If your dispute concerns pre-2014 conduct (e.g., a legacy supply or license relationship), the analysis shifts to district-court invalidity and license defenses, not PTAB. There is no AIA trial vehicle available to you: you cannot IPR an expired patent to obtain prospective relief in a meaningful way, CBM is closed to new petitions (barred for petitions filed on/after 2020-09-16), and PGR was never available for this pre-AIA patent.
3. Confirm there was no ex parte reexamination if you intend to rely on the prosecution record. My search did not surface one, but I could not rule it out from the sources available; the PTAB E2E / Patent Center "Application Data" tab (https://ptacts.uspto.gov/ptabweb) is the authoritative check. Ex parte reexam would not change the expiration date, but it could change the claim text you'd be reading.
4. Do not represent the absence of IPRs as an endorsement of validity. The correct framing for a client is: no AIA trial was ever filed against US 6,600,996, so no claim has been tested in a post-grant forum — and the patent expired 2014-10-21, so the question is academic. Well-asserted patents do eventually attract IPRs; this one attracted none, consistent with a term that lapsed before the AIA trial era matured.

Confidence note: I have high confidence in the "no PTAB activity" conclusion because the canonical ODP block says so and targeted searches corroborate it. I have not independently verified the maintenance-fee and expiration entries beyond the Google Patents/ODP legal-status rendering — if the stakes warrant, pull the Patent Center maintenance-fee record directly before relying on expiration in correspondence.

I'll research the assignment record for US 6,600,996, focusing on USPTO Assignment Center records, reel/frame numbers, and correspondents.

Let me try to pin down the specific reel/frame records and correspondents.

Let me search for patent-specific assignment records and the recording correspondents.

I hit my research step limit before I could open every '996-specific record directly in Assignment Center. Here is what I could verify, with unverifiable items flagged rather than filled in.

Inventors

Pattern worth noting: There is no fire-sale-style inventor exodus signature here. What is unusual is a two-batch confirmatory assignment of the same seven inventors' rights to the same assignee — recorded 2000-09-18 and again 2002-01-14. That is a records-perfection pattern (e.g., the original assignment from the 1994 priority filing not having been recorded against the 1997 CIP application number), not an ownership change. I am not able to confirm the precise cause without the underlying assignment documents.

Original assignee

Affymetrix, Inc. — original assignee of record and, per Google Patents, still the current assignee.

• Corporate form: Originally a California corporation; reincorporated in Delaware effective 1998-09-28 (shareholder vote June 1998; the merger record on the issued patent is the CA→DE change).
• Product embodying the claims: Yes. The independent claims (1, 7, 14) cover PM/MM expression monitoring; Affymetrix commercially shipped GeneChip® expression arrays and the accompanying analysis software (the specification itself describes the GeneChip/scan/analysis pipeline at FIGS. 3–4 and 18–19). Product references appear in Affymetrix customer agreements in later SEC filings (e.g., GeneChip® System 3000Dx, Human Genome U133 Plus 2.0 arrays).
• Primary line of business: DNA/oligonucleotide microarray ("GeneChip") manufacture and instrumentation; a classic operating company, not a licensing vehicle.
• Current status: Acquired — operating as a subsidiary. Affymetrix was acquired by Thermo Fisher Scientific in January 2016 (footnote in Nature Biotechnology, Vol. 35 No. 6, June 2017: "In January 2016, Affymetrix was acquired by Thermo Fischer Scientific..."). Note the sequencing quirk: US 6,600,996 expired 2014-10-21, i.e. before the Thermo Fisher acquisition, and no assignment of '996 to Thermo Fisher appears in the Google Patents legal events — consistent with the patent having lapsed out of the portfolio first.

Assignment timeline

Sources: Google Patents legal events for US 6,600,996 (in the authoritative full text supplied) plus Espacenet INPADOC records for sibling Affymetrix patents to recover reel/frame prefixes. I could not open the '996-specific Assignment Center entries, so per-record reel/frame and correspondent fields below are marked unconfirmed where I did not directly observe them.

1. 1998-09-28 (effective) / recorded 1999-09-07 (Google Patents) — Reel/Frame not confirmed for '996

   • Conveyance: Merger
   • Assignor: Affymetrix, Inc. (a California corporation)
   • Assignee: Affymetrix, Inc. (a Delaware corporation)
   • Correspondent: not retrievable in my searches. For sibling Affymetrix patents the same merger was recorded via Affymetrix's in-house legal department (the legacy-assignments PDF for Reel 014171 shows filings "From-Affymetrix Legal"); I could not confirm the '996 correspondent.
   • Context: internal reorg / change of domicile — Delaware reincorporation, not a sale.
   • Reel/frame context: sibling patents US 6,491,871 and US 6,410,229 carry this merger at Reel 010215 (frames 0164 and 0204 respectively); a different Affymetrix merger recording at Reel 009662/0497 also exists for US 5,593,839. The '996 frame is not confirmed — I am deliberately not guessing it.

2. 2000-09-18 — Reel/Frame not confirmed

   • Conveyance: Assignment of assignors' interest
   • Assignors: Jevons, Luis C.; Bernhart, Derek; Mittmann, Michael P.; Webster, Teresa A.; Morris, MacDonald S.; Lockhart, David J.; Ho, Ming-Hsiu (all seven)
   • Assignee: Affymetrix, Inc.
   • Correspondent: not retrievable
   • Context: records perfection — inventors' rights formally assigned to the employer.

3. 2002-01-14 — Reel/Frame not confirmed

   • Conveyance: Assignment of assignors' interest
   • Assignors: same seven inventors
   • Assignee: Affymetrix, Inc.
   • Correspondent: not retrievable
   • Context: confirmatory / duplicate recording — near-identical recording two years later; again a records-cleanup signature, not a transfer of ownership to a third party.

4. 2012-06-25 (executed) / recorded 2012-06-27 — Reel/Frame 028465/0541 (reel/frame confirmed on sibling Affymetrix patents; not directly opened for '996)

   • Conveyance: Security Agreement
   • Assignor: Affymetrix, Inc.
   • Assignee: General Electric Capital Corporation, as Agent
   • Correspondent: not retrievable
   • Context: securitization — collateral grant under the $100,000,000 Credit Facility of June 25, 2012 among Affymetrix as Borrower, GE Capital as Agent, Silicon Valley Bank as Syndication Agent (Credit Agreement text via Justia contracts). Not a transfer of title.
   • Cross-check: Espacenet INPADOC shows REEL/FRAME 028465/0541 ("SECURITY AGREEMENT; ASSIGNOR: AFFYMETRIX, INC.") on US 6,491,871, US 6,308,170 and US 5,593,839 — i.e., a single portfolio-wide recording, which is why the same reel/frame appears across the Affymetrix family.

5. 2015-11-13 — Reel/Frame not confirmed

   • Conveyance: Release by Secured Party
   • Assignor: General Electric Capital Corporation, as Agent
   • Assignee: Affymetrix, Inc.
   • Correspondent: not retrievable
   • Context: lien release — the 2012 security interest was discharged (facility repaid/refinanced). Title returns unencumbered to Affymetrix.

If Assignment Center returns no '996-specific records beyond the entries above, that is itself the finding: the only post-issuance "transfer" of record is a lien, not an ownership change, and it was released. No NPE ever appears in the chain.

Timeline diagram

timeline
    title Ownership of US 6600996
    1994 : Priority application filed
    1997 : CIP application filed
    1998 : Affymetrix reincorporates in Delaware
    1999 : Merger recorded California to Delaware
    2000 : Inventors assign rights to Affymetrix
    2002 : Confirmatory inventor assignment recorded
    2003 : Patent issued
    2012 : Security interest granted to GE Capital
    2014 : Patent expires
    2015 : GE Capital releases security interest

NPE / troll-pattern signals

1. Shell-entity transfer — not present. No "IP/Holdings/Ventures" LLC appears. Chain is Affymetrix → Affymetrix (merger) → Affymetrix → GE Capital as collateral agent → release back to Affymetrix. GE Capital is a commercial lender, not a patent-holding vehicle.
2. Known asserter in the chain — not present. Neither Affymetrix nor General Electric Capital Corporation appears on the Acacia / Marathon / IV / IPNav / Wi-LAN / Conversant / Vringo / Pendrell / Round Rock lists. Affymetrix is a product manufacturer.
3. Repeat correspondent across the chain — unclear. The Assignment Center correspondent fields (the single most probative tell per the brief) were not retrievable in my searches. I cannot confirm or rule out a repeat recording attorney. This is the one gap I would close first.
4. Cascading transfers — not present. No consecutive LLC-to-LLC assignments; the 2000 and 2002 filings are same-party perfection records to the same assignee, not transfers.
5. Pre-litigation transfer — not present. No infringement suit naming US 6,600,996 was found in my earlier litigation pass, so there is no pre-suit assignment to time.
6. Bankruptcy fire-sale — not present. Affymetrix was acquired by Thermo Fisher as a going concern in 2016; there is no Chapter 7/11 or patent auction in the record. (The '996 patent had already expired in 2014.)
7. Privateering — not present. No operating-company → NPE transfer; the only outbound conveyance (the GE Capital lien) is a financing instrument, not a competitive-assertion vehicle.
8. Defensive aggregator — not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN; it terminates back at Affymetrix.

Verdict

Operating-company assertion — specifically, operating-company ownership/retention with no NPE involvement.

Justification: the entire recorded chain (Google Patents legal events 1999-09-07 merger; 2000-09-18 and 2002-01-14 inventor assignments; 2012-06-27 security agreement at Reel 028465/0541; 2015-11-13 release) begins and ends with Affymetrix, Inc., a GeneChip microarray manufacturer that shipped products embodying the PM/MM expression-monitoring claims and was a documented serial enforcer of its microarray portfolio (Synteni/Incyte 1998; Hyseq 1998; Illumina 2004). Every event is either an internal reorg, an inventor-to-employer perfection filing, or a lien/release — none is an NPE transfer. Caveat: the verdict reflects ownership structure, not confirmed assertion of this patent — no suit naming US 6,600,996 was found, and the patent expired 2014-10-21.

Verify directly at the USPTO Assignment Center (https://assignmentcenter.uspto.gov/) by searching patent number 6600996; confirm the reel/frame and, especially, the correspondent of record on each entry, which was the one field I could not retrieve.

I'll research the specific prior-art references cited on the face of US 6,600,996.

Let me pull the actual front-page reference list for US 6,600,996 specifically.

Let me try to retrieve the actual front-page "References Cited" list from the patent document itself.

Prior-Art Analysis — US 6,600,996 B2 ("Computer-aided techniques for analyzing biological sequences")

Scope and reliability note (read first)

I need to be candid about what I could and could not establish, because it materially affects the §102 answer.

Limitation 1 — the authoritative front page is missing from the source I was given. The full text supplied is the description + abstract rendering; it does not contain the front-page "(56) References Cited" block (U.S. Patent Documents / Foreign Patent Documents / Other Publications). That block is the only place an examiner- or applicant-cited prior-art list appears. I attempted to retrieve it directly (Google Patents, FreePatentsOnline, patentimages front page, USPTO) but my search tooling was exhausted before I could confirm it.

Limitation 2 — a citation table I did retrieve does NOT belong to this patent. A Google Patents citation table surfaced in the searches ("Patent Citations (6)" + "Non-Patent Citations (4)") containing:

plus NPL: Chee et al., Science 274:610-614 (1996-10-25); Wang et al., Science 280 (1998-05-15); Proudnikov, Biologicals 28:57-66 (2000-06).

That table appeared on the Google Patents page for EP 1 184 467 A3 ("Screening method for gene variation," Canon, priority 2000), where US 6,600,996 is listed only as a similar/citing-family document. The tell: two of the six (EP 0 967 291 A1, 1999; EP 0 995 804 A2, 2000) and two of the NPL items (Wang 1998; Proudnikov 2000) post-date this patent's 1997-03-28 filing and therefore cannot be prior art against it. I am not treating that table as US 6,600,996's citation list. Flagging this because it is exactly the kind of conflation the "do not auto-correct identifiers" rule is meant to prevent — it is a different document's citations, not '996's.

Limitation 3 — avoid the near-number trap. US 6,045,996 (also Affymetrix, a GeneChip patent) appears throughout the search results (e.g., cited in JP 4,997,113 B2). That is a different patent from US 6,600,996. I have kept the identifier as given.

Because of Limitation 1, the set below is grounded in the patent's own specification text (which is authoritative and was supplied verbatim), not in a verified front-page (56) block. I label confidence accordingly. Treat the §102 conclusions as analytical, not as a restated examiner rejection.

§102 framework that governs this patent (pre-AIA)

US 6,600,996 was filed 1997-03-28 with earliest priority 1994-10-21 (through US 08/327,525 → US 5,795,716), and is a CIP of US 08/529,115 (→ US 6,040,138, filed 1995-09-15) and US 08/531,137 (→ US 5,974,164). Because it was filed before 2013-03-16 and has a pre-AIA effective date, pre-AIA 35 U.S.C. §102 applies:

• §102(a) — reference must predate the applicant's invention date.
• §102(b) — reference published/patented more than one year before the relevant filing date (i.e., on/before 1996-03-28 for the 1997 filing; ~1994-09-15 if the expression subject matter is carried back to the 1995-09-15 parent). Note that an applicant's own earlier publication counts as a §102(b) bar.
• §102(e) — US patent or published application (or PCT designating the US and published in English) with an effective date before the applicant's date.
• §102(g) — prior invention by another.

Critical filter for this case: The issued independent claims (1, 7, 14) are all directed to gene expression — PM/MM probe pairs tiled across a gene, comparing Ipm vs. Imm, and emitting an expression call (P/M/A) or a baseline-vs-experiment change call. Many of the cited documents are about base calling / sequencing, which is a different statutory subject matter from the issued claims. That distinction drives most of the "no anticipation" conclusions below.

Building on the prior sections' caveat: the verbatim claim text used here still rests on the secondary rendering, so claim-by-claim anticipation cannot be stated with full confidence.

References cited within the specification (authoritative — from supplied full text)

These are the applicant's own "incorporated by reference" citations in the background/description. On Google Patents these typically also populate the "Patent Citations" table. Dates in bold are ones I hold with high confidence; others are flagged.

Analytical takeaway from Category A. None of the array-synthesis or base-calling references (A1–A8, A10) discloses the full combination recited in independent claims 1, 7, or 14 (PM/MM pair comparison across a gene → expression call → optional baseline change display). They are §102(b)/(e) art but do not anticipate those claims. US 6,040,138 (A9) is the one document in this category whose disclosure overlaps the expression-monitoring claims.

The double-patenting / §102(e) problem with US 6,040,138 (A9)

US 6,600,996 is a continuation-in-part of US 08/529,115 (→ US 6,040,138). That creates a specific doctrinal situation:

• The '138 patent has an effective §102(e) date of 1995-09-15, before the '996 invention date. If its disclosure describes "comparing hybridization intensities of each pair of perfect match and mismatch probes … generating a gene expression call," it is facially §102(e) art against claims 1/7/14.
• However, because '996 and '138 are **commonly owned and share the same inventive entity line (Lockhart et al.)**, the standard examiner treatment of a CIP-to-parent overlap is §102(e) / obviousness-type double patenting, not clean third-party anticipation. '996's expression claims draw their earliest support from the '138 parent, so the applicant can generally antedate or is met with ODP rather than §102 anticipation.
• Conclusion: A9 is the closest thing to anticipating art for claims 1, 7, 14 in this set, but its status as a same-family, commonly-owned parent makes it a double-patenting problem, not a classic §102 anticipation by another.

References I could not verify as being cited on this patent's face

I located several Affymetrix-family documents (US 5,800,992; US 5,837,832; WO 95/11995) and the Drmanac patent (US 5,202,231) among the family/similar-document listings for this technology, but I was unable to confirm that any of them appears in US 6,600,996's (56) block. I list the two most analytically important, with explicit "attribution unverified" status:

• WO 95/11995 (Affymax/Affymetrix, published 1995-05-04) — "Arrays of nucleic acid probes on biological chips." If cited, this is the most dangerous §102(b) reference: published ~23 months before the 1997-03-28 filing (>1 year, so a statutory bar even though it is the applicant's own work), and it discloses high-density tiled arrays including PM/MM probe pairs used to detect/quantify targets and monitor expression. It would be the reference to test against claims 1, 7, 14. Attribution unverified.
• US 5,202,231 (Drmanac, issued 1993-04-13) — "Method of sequencing of genomes by hybridization." §102(b) art; but it is sequencing-by-hybridization, not PM/MM expression monitoring, so it would not anticipate claims 1/7/14. Attribution unverified.

Direct answer to "which claims does each reference potentially anticipate?"

Because (a) the verified references are predominantly the applicant's own family/background art and (b) the issued independent claims are limited to gene expression:

Only a reference that discloses the full combination — (i) PM/MM pairs where each pair targets a different portion of the same gene, (ii) comparison of Ipm and Imm within each pair to (iii) output an expression call, and for claims 7/14 a baseline comparison — can anticipate. On the verified record, the only document that approaches this is US 6,040,138, and it is same-family art (double-patenting posture). The specification's other citations are §102(b)/(e) references that do not, on their face, anticipate the expression claims.

Bottom line and recommended verification

1. I could not retrieve the authoritative (56) "References Cited" block for US 6,600,996 from USPTO/Google Patents/patentimages within this session. The citation table a search returned (US 5,202,231; WO 95/11995; EP 0 717 113 A2; US 6,027,880; EP 0 967 291 A1; EP 0 995 804 A2) more plausibly belongs to EP 1 184 467 A3, not to '996 — two of its members post-date '996's 1997 filing. Do not rely on it as '996's art.
2. On the verified specification citations, no reference cleanly anticipates claims 1, 7, or 14. The only close document is the parent US 6,040,138 (same family → double-patenting/§102(e), not third-party anticipation). WO 95/11995 would be the key §102(b) reference if it is on the face.
3. To close this out definitively, pull the actual front page of US 6,600,996 (USPTO PatentCenter "References Cited" / the issued PDF's (56) block) and the file-wrapper IDS/PTOL-892 forms, then run each cited reference against the exact issued claim text. My search budget was exhausted before I could do that, and I will not fabricate the (56) list.

One contradiction to flag with the earlier sections: the prior summary states the claims are directed to gene expression (correct), while the specification's summary recites "hybridization infinity." As noted, that is a published-text artifact; it does not change the §102 analysis but should not be silently normalized if you quote the claims.

I'll ground this analysis in the actual prior-art references, then build the § 103 combinations.

Let me pull the actual prior-art citation lists for this patent and its family.

Obviousness Analysis — US 6,600,996 B2 under 35 U.S.C. § 103

Analyst note on scope. The patent's effective filing date is 1997-03-28 (App. No. 08/828,952), with an earliest claimed priority of 1994-10-21. Because the application has an effective filing date well before 2013-03-16, the pre-AIA § 103(a) applies, and the prior-art categories are the pre-AIA § 102(a)/(b)/(e)/(g) buckets. I am working from the Google Patents "Prior art keywords" block (gene, probes, nucleic acid, base, perfect match), the references cited in the '996 specification, and the references the EPO examiner cited against the European counterpart EP 0 848 067 — all of which I retrieved and quote below. I flag the limits of my evidence at the end.

0. Two threshold matters that must be settled before the § 103 merits

(a) The § 120 benefit question is the whole ballgame. The '996 specification states it is a continuation-in-part of Ser. No. 08/529,115, filed Sep. 15, 1995 (now US 6,040,138) and of Ser. No. 08/531,137 (now US 5,974,164). Whether the gene-expression claims get the 1995-09-15 (or 1994-10-21) date controls which art exists:

Because the '138 alleges PM/MM expression monitoring with claims covering exactly that (see § 1 below), an examiner or challenger would likely have to concede § 120 benefit for the expression claims — which removes the single best reference. That same family relationship instead raises a non-statutory/statutory double-patenting objection (same owner, same invention as the '138), which is a different and generally weaker line. I give both branches below.

(b) A contradiction in the earlier-generated sections that I am flagging as instructed. The earlier "Patent summary" states the independent claims are 1, 7 and 14, all directed to gene expression, with no base-calling independent claim. The earlier "Strategic summary" states "every claim — 1–19 (base-calling/display claims) and the gene-expression monitoring claims 6, 13, 14 and dependents." Those two statements cannot both be right. I note it because it changes the § 103 target: if any independent claim covers base calling, it is anticipated in substance by US 5,795,716 and US 5,974,164 (its own § 120 parents, both naming Chee as an inventor and both already claiming the display-and-compare apparatus). I analyze the gene-expression independent claims as the live target and treat the base-calling claims conditionally.

1. The prior-art universe (with what each reference actually teaches)

2. Claim 1 — element-by-element

(Claim 1, as reported by the secondary sources in the earlier section: a computer-implemented method comprising inputting hybridization intensities of a plurality of PM/MM probe pairs, the PMs being perfectly complementary to the gene and the MMs having at least one base mismatch, with each pair targeting a different portion of the gene, the intensities being indicative of gene expression; comparing the intensities of each pair; and generating a gene expression call.)

Result: every element of claim 1 is disclosed in a single reference (Lockhart 1996 and US 6,040,138 each independently), and certainly in the combination. If Lockhart 1996 is available, claim 1 is not merely obvious — it is close to anticipated under § 102(a). The only reason it is not a clean § 102 case is that the reference measures a level rather than emitting a categorical call; that gap is closed by obviousness (see § 5).

3. The § 103 combinations

Combination A (primary): Lockhart 1996 + US 6,040,138 — claim 1

Proposed rejection: Claims 1–6 obvious over Lockhart et al. (Nat. Biotechnol. Dec. 1996) in view of US 6,040,138 and US 5,795,716.

• Lockhart 1996 supplies the assay architecture (PM/MM pairs across a message; MM as internal specificity control; PM/MM analysis to recognize low-abundance transcripts) and the word-for-word claim vocabulary.
• US 6,040,138 supplies the computation the claim requires — the difference and average difference between each PM and its corresponding MM, aggregated per gene — and the probe-count/density context (>100 probes; >60/cm²; ≥10 probes per gene).
• US 5,795,716 supplies the computer element (analysis system 126) and the display/comparison framework.

Motivation (KSR factors, 550 U.S. 398 (2007)):

1. Same field, same problem, same art. Both references address quantitative, parallel measurement of mRNA abundance from a single biological sample on a high-density array. The problem Lockhart 1996 states ("low-intensity hybridization patterns from rare RNAs … accurately recognized in the presence of cross-hybridization signals") is the same problem the '996 addresses in FIG. 19 (NPOS/NNEG counting against a background).
2. The references expressly point at each other. The '996 specification itself lists US 5,795,716, US 5,733,729 and US 5,974,164 as the computer-program products "for probe design, management of data, analysis, and instrument operation" used with the same GeneChip arrays (this listing recurs verbatim across the Affymetrix family, e.g. US 2008/0261817 A1, ¶[0025]). That is an express teaching that the array chemistry and the analyzing software are designed to be used together.
3. Common ownership / same research program. All of the primary references are Affymetrix (or Affymax, Affymetrix's predecessor) documents. Under MPEP 706.02(l) and KSR, a common assignee and a shared development program is strong evidence of a motivation to combine.
4. Predictable result. Implementing a scoring rule that differences a probe against its own single-base mismatch control is within the ordinary skill of a bioinformatics programmer; Lockhart 1996 already states the PM/MM analysis is what confers specificity, so the only remaining step is arithmetic.
5. Thresholds are disclosed as arbitrary. The '996 specification itself says the difference and ratio thresholds "are typically user defined values that have been determined to produce accurate expression monitoring" (D = 20, R = 1.2). Where the patentee concedes a parameter is a tuning choice, KSR deference to the artisan applies.

Combination B: WO 95/11995 + US 5,795,716 (+ US 5,733,729) — the fallback if § 120 benefit is granted

If the expression claims take the 1995-09-15 date, Combination A's references (all post-dating that date except the '716/'729) shrink. The fallback:

• WO 95/11995 (Fodor/Affymax) — high-density probe arrays with matched/control probes for interrogating a target (the EPO examiner cited it against every claim of the counterpart, "page 30, line 3–line 33; claims 37–41," category A).
• US 5,795,716 (Chee, 1994) — computer analysis of hybridization intensities plus matched vs. mismatched probe sets, and side-by-side display of reference and sample sequences / comparative analysis of multiple experiments.
• US 5,733,729 (Lipshutz/Walker, Sept. 1995) — unit/cell array layout and probe-binding analysis.

Motivation: all three are directed to "reading" a light-directed probe array by computer; the '716 and '729 are already acknowledged in the '996's own background as the base-calling precursors ("Innovative computer-aided techniques for base calling are disclosed in U.S. patent application Ser. No. 08/531,137 … Ser. No. 08/528,656 … Ser. No. 08/618,834"). The step from "call bases from PM/MM intensity patterns" to "call expression from PM/MM intensity patterns" is a reapplication of the same algorithm to a different array layout — the classic KSR "known technique, ready for improvement" situation.

Combination C: US 6,040,138 / Lockhart 1996 + Schena 1995 — for the differential claims

For claims 7 and 14, add Schena et al., Science 270:467–470 (1995) (two-sample, differential, quantitative expression monitoring with a reference standard) to supply the baseline-comparison step, and US 5,795,716 to supply the display.

4. Claims 7 and 14 — baseline comparison and display

Claim 7 adds: (i) generating a gene expression level; (ii) comparing it to a baseline expression level; (iii) determining a change; and (iv) displaying the change.

Claim 14 is claim 7 minus the display step and is therefore broader than claim 7; if claim 7 is obvious, claim 14 is a fortiori obvious (it removes a limitation rather than adding one).

5. The "expression call" (present / marginal / absent) and the numeric limits

The only limitation with no express antecedent in the strongest references is the categorical call ("expressed, marginal, or absent") and the specific numbers.

• Transforming a continuous measurement into a thresholded category is not an inventive act. The '996 itself converts a numerical ratio into a call by comparing it to R = 1.2 (FIG. 11) and into P/M/A states by binning P-values (FIG. 19). The strongest evidence of obviousness is that the same value, 1.2, appears in both the base-calling ratio cutoff and the expression ratio threshold — the patentee re-used a parameter from its own admitted prior art base-calling algorithm. That is a strong § 103 position on any dependent claim reciting R ≈ 1.2.
• Background subtraction (subtracting the "blank" cell intensity, as in Claim 1's dependent claims) is disclosed on the face of the '996 as conventional for blank cells on the same arrays (FIG. 5) and is inherent in the '138's "mismatch control" rationale.
• Difference threshold D = 20 is a tuning value that the specification does not show to be critical. Absent evidence that 20 produces a result not otherwise obtainable, a recited numeric threshold is an obvious design choice (In re Woodruff line of authority; MPEP 2144.05).

6. The strongest counter-arguments a patent owner will make (and my assessment)

1. "The references are all our own family — they are not § 102/§ 103 art." This is the best defense and it is doctrinally serious. US 6,040,138 is the direct § 120 parent of the '996; if the expression claims are supported by 08/529,115, the '138 is not prior art at all, and Lockhart 1996 (published after the parent's filing) is likewise out. Assessment: the defense succeeds only if § 120 benefit is established claim-by-claim. The '996's expression disclosure adds the decision matrix, NPOS/NNEG, NINC/NDEC, and the D/R/DDIF/RDIF thresholds, which are not in the '138's claims. If the claims are construed to require the decision matrix (as the FIG. 19/26 disclosure suggests), the patentee may have to concede that those limitations are new matter as of 1997-03-28 — which re-opens Lockhart 1996 and Chee 1996 as art. This is a genuine § 120 trap, and it is the central issue in any validity challenge.
2. "Non-analogous art / different field." Weak here. Both references are microarray-based nucleic acid analysis; the fields are identical.
3. "No reasonable expectation of success." Weak. Lockhart 1996 reports 1:300,000 sensitivity and three orders of magnitude of linearity — i.e., it reports success.
4. Secondary considerations. Affymetrix's GeneChip commercial success is real, but the nexus to these claims is weak: the '138 and the '716/'729 patents (and the underlying VLSIPS chemistry) are at least equally responsible, and the '996 is a CIP that bundled in the base-calling material from its parents. Commercial success attributable to the platform rather than to the claimed expression-call algorithm will not rebut a strong prima facie case.

7. Bottom line

• If the '996's expression claims reach back to 1995-09-15 (§ 120), the most powerful art disappears and the § 103 case must be built on WO 95/11995 + US 5,795,716 + US 5,733,729 (Combination B) — a workable but contestable case, and one the EPO examiner considered but did not treat as anticipating.
• If the expression claims are dated 1997-03-28 (i.e., the decision matrix and thresholds are CIP new matter), then Lockhart et al. (Nat. Biotechnol., Dec. 1996) alone or in view of US 6,040,138 renders claim 1 obvious (Combination A), with Schena 1995 closing the baseline step for claims 7 and 14. Claim 1 in that posture is arguably anticipated, because Lockhart 1996 uses the claim's exact vocabulary ("perfect match," "mismatch," "probe pair," "subsequence," "PM/MM analysis") and applies it across many subsequences of one message.
• Dependent claims on thresholds (D, R, P/M/A) and on graphical display are the most vulnerable and the easiest to dispose of on KSR grounds, because the patentee's own specification characterizes the thresholds as "user defined values" and re-uses R = 1.2 from its admitted base-calling prior art.
• Practical relevance: none, for infringement purposes. As established in the prior sections, the patent's anticipated expiration was 2014-10-21 and it stands "Expired – Fee Related." A § 103 analysis of a dead patent is an academic exercise unless the goal is to evaluate the family (US 6,040,138; US 5,795,716; US 5,733,729; US 5,974,164) or to bolster a collateral estoppel/§ 282 defense for pre-2014 conduct.

8. Evidence caveats (please read before relying on the above)

1. I do not have the verbatim issued claim text. The earlier sections drew claim language from RPX/Google Patents reference listings and noted the source ellipses. Every element mapping above should be re-run against the USPTO PatentCenter/Patent Cooperation Treaty copy of the '996 claims before filing anything. The element charts are only as good as the claim text.
2. The "hybridization infinity" anomaly — now corroborated as an artifact. As I noted previously, the US text reads "hybridization infinity" in the summary passages. The EP 0 848 067 A2 parallel text reads "indicating hybridization intensity between the perfect match and mismatch probes" (EP 0 848 067 A2, Summary). I am not auto-correcting the identifier or the quoted text; I am reporting that the two renderings diverge, which is a prosecution-history point worth checking.
3. I did not retrieve the Google Patents "Patent Citations" (front-page 56) list for the '996 itself. I inferred the cited-art set from the specification's incorporated-by-reference list and the EPO search report on the counterpart. If the '996's own Form 892 lists additional U.S. references (e.g., US 5,445,934; US 5,800,992; Drmanac-type sequencing-by-hybridization art), those should be added and may strengthen the case — particularly Drmanac-style SBH art, which would be useful against any independent base-calling claim.
4. Priority/§ 120 benefit is unresolved and is dispositive. I have flagged it rather than resolved it, because resolving it requires the 08/529,115 specification as filed (to test written-description support for the decision matrix) and the 08/531,137 record.

Sources cited: US 6,600,996 B2 · US 6,040,138 claims · Lockhart et al., Nat. Biotechnol. 14:1675 (1996) · US 5,795,716 biblio · US 5,974,164 · US 6,228,593 (continuation of 5,733,729) · EP 0 848 067 A2 · EP 0 848 067 A3 search report · US 2008/0261817 A1 (Affymetrix software-program listing).