US 5795716

I'll search for authoritative information on US 5795716, including the USPTO record and any CAFC/litigation activity.

Let me check the claims in full, and look for litigation/CAFC activity.

US Patent 5,795,716 — Analyst Summary

Note on scope of this memo: I retrieved the patent's own text and the Google Patents, Justia, OSTI, and Espacenet records. My web searches did not return any USPTO or Federal Circuit 2026 docket activity for this patent, and I could not retrieve the complete, numbered claim set from an authoritative source. Where I lack authoritative data, I say so explicitly rather than infer.

---

1. Bibliographic data (from the patent text and Google Patents/OSTI/Espacenet records)

Caveat on the Delaware Docket: Google Patents attributes case 1:04-cv-00901 to the family ("Family has litigation"), not necessarily to the '716 patent itself, and I could not independently confirm from a court record that 5,795,716 was an asserted patent in that case. Treat the litigation link as an unverified lead.

Caveat on inventor lists: Several later sources (e.g., the FPO/Justia "Chee et al." citations, and the continuation US 6,733,964) list a five-inventor team: Chee, Chunwei Wang, Luis C. Jevons, Derek H. Bernhart, and Robert J. Lipshutz. That team appears on the continuation patents, not on the '716 face. The '716 assignment recorded 1995‑08‑04 names those same five as assignors.

---

2. Abstract (verbatim)

"A computer system for analyzing nucleic acid sequences is provided. The computer system is used to perform multiple methods for determining unknown bases by analyzing the fluorescence intensities of hybridized nucleic acid probes. The results of individual experiments are improved by processing nucleic acid sequences together. Comparative analysis of multiple experiments is also provided by displaying reference sequences in one area and sample sequences in another area on a display device."

---

3. Plain-language overview

The patent is a software/data-analysis patent sitting on top of Affymetrix's light-directed DNA array ("VLSIPS") chemistry. The chip itself, the probes, and the scanner are prior work (cross-referenced to U.S. Pat. No. 5,143,854 and application Ser. No. 08/249,188). What is claimed is the computer-implemented processing of the scanner's fluorescence intensity files to call bases.

The specification discloses four analytic engines:

1. Intensity Ratio Method (§II): Background-subtract the four "mutation" probe intensities, sort descending, then take successive ratios (base 1:base 2, then 2:3, then 3:4) and compare each to a cutoff. If a ratio clears the cutoff, call the complement of the top base; otherwise fall through to IUPAC ambiguity codes (N = insufficient intensity, X = insufficient discrimination).
2. Reference Method (§III), two implementations: Compare a sample's probe intensities against a known reference sequence's probes. Version one uses ratios-of-ratios per base; version two normalizes each base's intensity to the sum of all four bases, subtracts the reference's normalized ratios from the sample's, and looks for differences exceeding upper/lower cutoffs, including a characteristic neighboring-position difference pattern (FIG. 5C/5D). This is the more robust method under cross-hybridization.
3. Statistical Method (§IV): Build means and standard deviations for wild-type and "other" base ratios across many reference experiments, then score the sample's ratios against per-base standard-deviation cutoffs. A preprocessing stage (Microfiche Appendices C and D) caches the reference statistics.
4. Pooling Processing (§V): Label reference and sample nucleic acids with different fluorophores (e.g., fluorescein ~530 nm and rhodamine ~580 nm), combine them, and hybridize simultaneously on the same chip so that most experiment-to-experiment variation is shared. This is the direct ancestor of two-color microarray hybridization.

Added to that is VIEWSEQ™, a comparative visualization GUI (FIGS. 8–13) with a reference-sequence pane and a sample-sequence pane, sequence linking, color-coded mismatches, and per-base intensity-graph windows. C++ source for CALLSEQ™ and VIEWSEQ™ is supplied on Microfiche Appendices A–E (5 sheets, 272 frames). The specification is expressly subject to a copyright reservation.

---

4. The independent claims (plain language)

The '716 claims are drafted as computer program product and system claims rather than as a method of sequencing — a notable drafting choice for a 1994 filing.

Claim 1 — "intensity comparison" computer program product (broadest):
A computer program product for identifying an unknown base in a sample nucleic acid sequence, with:

• code that receives signals for probe intensities of a plurality of nucleic acid probes, where each probe differs from the others by at least a single base, and each intensity reflects hybridization with at least one sequence including the sample;
• code that compares those probe intensities to each other;
• code that generates a base call from the result of that comparison and the known probe sequences; and
• a computer-readable medium storing the codes.

Claim 2 — "ratio" computer program product: Same framework, but the comparison is specifically a ratio of a higher probe intensity to a lower probe intensity, and the base call is the base of the probe having the higher intensity if that ratio exceeds a predetermined ratio value. (This is the Intensity Ratio Method in claim form.)

Claim 3 — "reference" computer program product: Two sets of codes receive a first set of probe intensities from hybridization with a reference nucleic acid sequence and a second set from the sample; code compares at least one intensity from each set; code generates a base call from that comparison and the probe sequence. (This is the Reference Method.)

Claim 4 — "statistical" computer program product: Codes receive statistics about a plurality of experiments (each producing probe intensities for a reference sequence), code compares the sample's probe intensities against those statistics, and code generates the base call. (This is the Statistical Method.)

Claims 5–8 — system claims: Four parallel independent claims in apparatus form — "a system that identifies an unknown base … comprising a processor; and a computer readable medium coupled to said processor for storing a computer program comprising …" — tracking the same four analytic approaches as claims 1–4, each ending in code that generates the base call.

Claims 9 and 10 — dependent system claims: Claim 9 recites that the plurality of nucleic acid probes are in an array of probes (depending on any of claims 5, 6, 7, or 8). Claim 10 recites that the probe intensities are fluorescent intensities (similarly depending on claims 5–8).

Google Patents' claim-text indexer records "computer program claims 8," consistent with claims 1–8 being independent, two in each of the four method families.

---

5. Explicit uncertainties

• Full claim set not verified. I retrieved the text of claims 1–10 only, and that text came from a Justia page whose excerpt was elided in places. I do not know with confidence the total number of claims in 5,795,716, nor whether claims 11 and higher exist. My searches did not surface a reissue, reexamination, or certificate of correction.
• Claim 1's "at least a single base" limitation is what distinguishes the '716 product claims from the narrower method claim 1 of the continuation US 6,733,964, which requires the probes to "differ from each other by a single base at an internal interrogation position." That '964 claim text I did verify directly from the Official Gazette (Issue 1282-2).
• No CAFC 2026 docket activity found. I searched for Federal Circuit 2026 dockets and for "5795716" in litigation contexts and found nothing. The only litigation trail located is (a) the Google Patents family flag for D. Del. 1:04-cv-00901, and (b) an Affymetrix, Inc. v. Illumina, Inc. opinion (via a vlex case-law page) discussing ownership/standing around a "'716 patent" and a 1991 Wagner–Affymax consulting agreement. I do not know with confidence that the "'716 patent" in that opinion is US 5,795,716 — the opinion ties "'716" to Dr. Lipshutz's conception, whereas the '716 patent names only Chee. There is a real risk it refers to a different patent ending in 716. Do not rely on that link without pulling the docket.
• Assignee-of-record nuance: OSTI and the original-assignee field show an individual (Chee) as assignee at grant, while Google Patents shows Affymetrix as current assignee. Both are true at different points in time; the chain runs Chee → Affymax Technologies N.V. → Affymetrix, Inc.

Family members (from the Google Patents legal-events timeline): US 08/531,137 (→ US 5,974,164); EP 95307476 (→ EP 0 717 113 A2); US 08/828,952 (→ US 6,600,996); US 09/049,805 (→ US 6,733,964); US 09/158,765 (→ US 6,242,180); US 09/796,071 (→ US 6,607,887); and later publications US 2003/0220748 A1 and US 2005/0186581 A1.

I'll research litigation involving US Patent 5,795,716. Let me search multiple sources.

I found several leads. Let me dig deeper into each case.

Let me get more specifics on the Delaware case and other potential suits.

Litigation involving US Patent 5,795,716

Interpretation note: Per your instruction I am treating the identifier literally. "US patent 5795716" = US 5,795,716 B1, "Computer-aided visualization and analysis system for sequence evaluation" (inventor Mark S. Chee; filed Oct. 21, 1994; issued Aug. 18, 1998; assignee Affymetrix, Inc.; now expired – lifetime, anticipated expiration Aug. 18, 2015). The Google Patents "Family has litigation" entry points to Delaware District Court case 1:04-cv-00901, which matches the first case below.

I located two district-court litigations in which the '716 patent was asserted, plus related USPTO reexamination activity. Details follow.

---

1. Affymetrix, Inc. v. Hyseq, Inc. (with Synteni, Inc. and Incyte Pharmaceuticals, Inc.)

Sources: CourtListener – Affymetrix, Inc. v. Hyseq, Inc. (docket numbers, citations, '716 discussion); Casetext – Affymetrix, Inc. v. Hyseq, Inc. ('716 accused claims); Hyseq 10-K reporting Oct. 2001 settlement

---

2. Affymetrix, Inc. v. Illumina, Inc.

Sources: Stipulated Order of Dismissal, D. Del. 1:04-cv-00901-JJF, Doc. 471-2 (Jan. 16, 2008); Markman order press release listing '716 as patent-in-suit (Aug. 17, 2006); CourtListener – Affymetrix, Inc. v. Illumina, Inc. (ownership standing ruling); vLex – Affymetrix, Inc. v. Illumina, Inc.; GenomeWeb – USPTO to Re-examine Two More Patents in Affymetrix, Illumina Litigation; IBO – Affymetrix Patents To Be Reexamined

---

3. Related USPTO reexamination (not a court case, but part of the '716 enforcement history)

At Illumina's request, the USPTO granted reexamination of the '716 patent (along with U.S. 5,545,531, 6,355,432, and 6,646,243) during the Affymetrix–Illumina litigation. Affymetrix reportedly sought reexamination of all claims. This was an administrative validity challenge, not separate litigation.

---

Caveats and limitations

• I did not find any other district-court, ITC, or CAFC litigation asserting the '716 patent. In particular, the '716 patent is not among the patents in the Zond v. Renesas complaint that surfaced in search results (that complaint references a different "'716 patent"), and it is not the MicroUnity v. AMD patents — both are false-positive hits I am flagging and excluding per your instruction not to conflate similar numbers.
• Exact filing dates for the Hyseq action and definitive settlement terms for the Illumina action could not be verified from the sources retrieved. I was unable to complete a full PACER/Unified Patents docket pull because the research session hit its tool-call limit; you may wish to confirm the Hyseq filing date and the terms/date of the Illumina settlement via PACER for C 99-21163 JF (N.D. Cal.) and 1:04-cv-00901-JJF (D. Del.), and via the Unified Patents litigation record at the Delaware case link cited on the patent's Google Patents page (https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A04-cv-00901).

Summary: Two known litigations asserted US 5,795,716 — Affymetrix v. Hyseq/Synteni/Incyte (N.D. Cal., C 99-21163 JF & C 99-21164 JF; settled Oct. 2001) and Affymetrix v. Illumina (D. Del., 1:04-cv-00901-JJF; filed July 2004; jury verdict for Affymetrix in 2007; dismissed with prejudice Jan. 2008) — plus Illumina-requested USPTO reexamination of the patent.

I'll start by verifying the structured data against live sources.

Let me probe a few more angles to be sure nothing is missed.

PTAB proceedings overview

Total AIA trial proceedings on file for US 5795716: ZERO. The structured "PTAB proceedings on file" block — sourced from the USPTO Open Data Portal — lists no IPR, PGR, or CBM proceedings for this patent, and my independent web searches surfaced no AIA trial petitions, institution decisions, or Final Written Decisions naming 5,795,716. The breakdown is therefore: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0. The bottom-line defensive posture is not "the patent is hardened by surviving IPRs" — it is that the patent has never been tested in an AIA trial at all, and separately that it expired on 2015-08-18, so the practical defense is expiry rather than an IPR-based invalidity record. Because no petition was ever filed, no § 315(e)(2) estoppel attaches to anyone, and a defendant today retains the full universe of prior-art grounds.

There are consequently no per-proceeding entries to report. Instead, here is what a defendant actually needs to know about the non-AIA challenges that do exist on the record — these are frequently confused with PTAB activity and should not be described as IPRs.

No AIA proceeding — but two non-AIA challenges are on the record

1. Ex parte reexamination (not an AIA trial) — requested by Illumina, ~2007

• Type: Ex parte reexamination under 35 U.S.C. §§ 301–307 / 37 C.F.R. § 1.510. This is a Central Reexamination Unit (CRU) proceeding, not a PTAB trial; it produces a reexamination certificate, not an FWD.
• Filed: Reported as requested on 2007-10-19, per BioCentury's report that Illumina asked the PTO to reexamine U.S. Patents 5,545,531; 5,795,716; 6,355,432; 6,646,243, and 6,399,365. (BioCentury)
• Institution: The USPTO ordered reexamination of 5,795,716. (GenomeWeb, "USPTO to Re-examine Two More Patents in Affymetrix, Illumina Litigation"; IBO). Note the reported date ambiguity in the IBO piece — the article is dated 12/2007 with an OCR-garbled year string ("12/21/0—-"), so treat the exact announcement date as uncertain.
• Outcome: I could not verify the disposition of this reexamination with high confidence. I found no reexamination certificate confirming or canceling specific claims in the sources available to me. Google Patents lists the legal status only as "Expired – Lifetime" with no claim-cancellation annotation, and the patent's front page as fetched shows no reexamination-certificate link. Do not represent to a court or an adversary that any claim of 5,795,716 was canceled in reexamination — I have not confirmed that, and I will not guess.
• Defensive value: A 2007 CRU proceeding on a patent that expired in 2015 is of historical interest only. It generated no PTAB record, no estoppel, and no citable Final Written Decision.

2. District court litigation (not a PTAB proceeding)

• Affymetrix, Inc. v. Illumina, Inc., D. Del. 1:04-cv-00901 (filed July 2004) — the Delaware case flagged in the structured data. This is the source of the reexamination requests and the venue of the March 2007 jury verdict reported in favor of Affymetrix. (Unified Patents litigation record)
• Affymetrix, Inc. v. Hyseq, Inc., N.D. Cal. (132 F. Supp. 2d 1212) — Affymetrix asserted claims 3, 4, 7, 8, 9, and 10 of the '716 patent; claims 3, 4, 7, and 8 are the independent claims. (CourtListener opinion). The decision also addresses ownership/standing and construe terms including "probe intensity," "probe," "indicating an extent of hybridization," and "corresponding to probe intensities for a plurality of nucleic acid probes" — a useful free claim-construction record if anyone asserts the patent again.
• Affymetrix, Inc. v. Illumina, Inc. standing/ownership decision (CourtListener): discloses that the '716 invention traces to Dr. Lipshutz's conception and that Affymax assigned its rights in the '716 application to Affymetrix in 1996 — i.e., the chain-of-title vulnerability that Illumina litigated (it lost, on the record available to me).
• Defensive value: These are the real-world challenges. Notably, the Hyseq case shows Affymetrix's own assertion set, which tells you exactly which claims a modern plaintiff would reach for — claims 3, 4, 7, 8, 9, 10 — and gives you a ready-made claim-construction record, plus a chain-of-title attack that has already been run once.

Strategic summary

Claim status: UNTESTED. No claim of 5795716 has been canceled, sustained, or construed by the PTAB, because no AIA trial was ever requested. There is no IPR-narrowed claim set, and there is no surviving-claim list to hand a licensing negotiator. The only public adjudicative narrowing came from district court claim construction (Hyseq, N.D. Cal.) and from any reexamination amendments, whose disposition I could not verify. Treat every claim of the patent as formally intact but expired.

Estoppel landscape: completely clean. There is no § 315(e)(2) estoppel, because there is no IPR. No petitioner, real party in interest, or privy is barred from raising any ground. Every § 102/§ 103 combination remains available in an IPR or in district court, and no IPR-based prior-art ground has been "used up." That said, the practical value of that clean slate is low: the patent's 20-year term ran from the 1994-10-21 filing date and Google Patents records anticipated expiration on 2015-08-18, so a petition filed today would be an expensive exercise about a patent with no prospective damages.

Pattern signals: the defensive-aggregator chain is empty. No entity — Unified Patents, RPX, or otherwise — has filed an AIA trial against 5,795,716. The structured data flags litigation family membership, and the family-level Darts-IP link (family 23276902) is present, but that reflects district court litigation, not PTAB litigation. The same patent owner (Affymetrix) was on the offensive side of that litigation and on the defensive side of the reexamination requests — a posture consistent with a well-asserted patent that nonetheless never attracted an IPR before it expired. That is a mild negative signal for a defendant: the patent survived its assertion era without an invalidity trial, so you have no precedent to lean on; but it is also unsurprising, because it expired at the moment IPR filing rates exploded.

Recommended next steps

1. Lead with expiry, not invalidity. US 5795716 expired (anticipated) on 2015-08-18. Any demand letter asserting it in 2026 can, at most, seek back damages for past conduct, and 35 U.S.C. § 286's six-year look-back (measuring from a complaint filed now, i.e., activity after roughly 2020-09-11) appears to fall entirely after expiration — meaning no recoverable damages window and no injunctive relief. Get an analysis of the actual expiration date (including any terminal disclaimer, PTA/PTE, or intervening laches/equitable estoppel issues) from litigation counsel; I flag this as a strategic observation, not a legal conclusion.
2. If anyone purports to rely on a PTAB outcome, demand the proceeding number. There are no IPR/PGR/CBM numbers to cite. If opposing counsel asserts that "the PTAB upheld" or "the Board invalidated" claims of 5,795,716, that representation is unsupported — the only post-grant challenge on the record is the 2007 ex parte reexamination, whose outcome I could not verify and whose certificate should be pulled directly from the file wrapper.
3. Pull the primary documents rather than relying on secondary press. Retrieve (a) the reexamination file wrapper and any reexamination certificate for 5,795,716 from USPTO Patent Center, and (b) the complete PTAB E2E record search at PTAB E2E using the patent number — the ODP "no proceedings" result should be independently confirmed there before you represent the absence to anyone.
4. Mine the Hyseq claim-construction opinion. If assertion is threatened, Affymetrix, Inc. v. Hyseq, Inc., 132 F. Supp. 2d 1212 (N.D. Cal. 2001) supplies constructions of "probe intensity," "probe," and "indicating an extent of hybridization," and identifies claims 3, 4, 7, 8, 9, 10 as the asserted set (3, 4, 7, 8 independent). That is your starting estoppel-free invalidity and non-infringement record.
5. No PTAB trial-stage deadlines to calendar. There is no pending proceeding, no institution-decision deadline, no oral hearing, and no statutory one-year FWD date to track. Do not populate a docket with phantom milestones.

---

Confidence / caveats: (a) The absence of PTAB proceedings comes from the structured ODP-sourced block in this prompt plus negative web searches — the latter are weaker evidence than a PTAB E2E patent-number query, which I recommend running directly. (b) I could not verify the outcome of the 2007 ex parte reexamination of 5,795,716, nor any appellate disposition of the Affymetrix/Hyseq or Affymetrix/Illumina district court cases; I have said so rather than inferring. (c) I have deliberately not supplied any proceeding number, judge panel, or FWD holding, because none exists on this record — inventing one would be worse than an empty section.

I'll research this patent's ownership chain across the sources available to me.

Research note on sourcing

I could retrieve the event-level assignment record for US 5,795,716 (via the Google Patents legal-events mirror of the USPTO assignment data, cross-checked against the Affymetrix v. Illumina docket and the Illumina 10-K). I could not retrieve reel/frame numbers or correspondents of record — the Assignment Center is behind a session-based search UI that my tooling cannot query directly, and no mirror I could reach exposes the reel/frame fields for this patent. I will not guess at those numbers. Where a field is unverified I say so explicitly below rather than filling it in.

---

Inventors

Five named inventors, confirmed by the inventor list of the identically-titled continuation US 6,733,964 (which claims priority to application 08/327,525, now US 5,795,716) and by the assignor list on the 1995 inventor→company assignment:

Pattern notes. No departure-within-12-months signal: the inventors did not scatter. All five executed the same assignment to Affymax Technologies N.V. roughly nine months after filing, and none of the later records show any inventor retaining or reacquiring an interest. The interesting anomaly is not inventor turnover but the filing-to-assignment gap — the application was filed 1994-10-21 while title still sat with the individuals, and the inventor→company assignment was not executed until 1995-08-04. That gap is exactly what Illumina attacked on standing a decade later.

Ownership dispute worth flagging. The Delaware litigation record contains an unusual chain-of-title wrinkle: an inventor-side interest traceable to a 1991 consulting agreement between a "Wagner" and Affymax Research Institute, whose Section 2 ("Ownership of Work Product") was the vehicle by which Affymax claimed the invention. Illumina argued that agreement did not effect a present assignment; the court held that it did, that "upon conception, legal title to the invention was transferred to Affymax by operation of law," and that "Affymax effectively assigned its rights in the invention to Affymetrix." I flag this because the "Wagner" name appears in the court record but not on the patent's face, and I could not resolve who Wagner was from the sources available. Treat as an unverified anomaly, not a finding.

Government rights (not an ownership transfer). Per the specification, portions arose under DOE contract 92ER81275 (SBIR) and NIH contracts H600813-1/-2. These create government license/march-in rights; they do not appear as recorded assignments.

---

Original assignee

None on the face of the patent — the patent issued to the individual inventor(s). Google Patents records the 1994-10-21 event as "Application filed by Individual" and lists "Original Assignee: Individual"; the DOE/OSTI record for this patent likewise lists the assignee as "Chee, Mark S., 3199 Waverly St., Palo Alto, CA 94306." Press coverage from the 2004–2007 litigation says the same: "the second patent is assigned to Mark Chee."

The commercial owner from 1996 onward was Affymetrix, Inc.

• Primary line of business: Affymetrix commercialized the VLSIPS/light-directed synthesis platform — GeneChip® oligonucleotide arrays, array scanners, and the accompanying analysis software. The claims here (computer-implemented base calling from probe intensities) read directly on Affymetrix's own instrument + software products, and the specification's software (CALLSEQ™, VIEWSEQ™) is Affymetrix product code.
• Did they ship a product embodying the claims? Yes. Affymetrix asserted this very patent against Illumina's arrays, scanners, and software in 2004, and the 2007 jury found infringement of five patents-in-suit including the '716 — a claim set that was commercially practiced by the plaintiff and by an actual competitor.
• Current status: the patent expired 2015-08-18 (anticipated expiration, per Google Patents; also recorded there as "Expired – Lifetime"). Affymetrix itself was acquired by Thermo Fisher Scientific, Inc. (announced Jan 2016, closed March 2016). Affymetrix was an operating company throughout — it never filed bankruptcy and never dissolved.

---

Assignment timeline

⚠️ Reel/frame and correspondent fields: not retrieved. I could not verify reel/frame numbers or correspondent-of-record names for any link in this chain, and I am not inventing them. The entries below are dated and typed from the legal-events record; the reel/frame line is left as [not retrieved] rather than fabricated. Anyone needing the reel/frame and attorney-of-record data should pull it directly from USPTO Assignment Center (search by patent number 5795716).

• 1994-10-21 (filed) / recorded n/a — Reel [not retrieved]

  • Conveyance: (none — original filing)
  • Assignor: n/a
  • Assignee: Individual inventors (Chee, Wang, Jevons, Bernhart, Lipshutz)
  • Correspondent: [not retrieved]
  • Context: Original filing; title remained with the inventors.

• 1995-08-04 (executed) / recorded [not retrieved] — Reel [not retrieved]

  • Conveyance: Assignment
  • Assignor: Bernhart, Derek H.; Chee, Mark S.; Jevons, Luis C.; Lipshutz, Robert J.; Wang, Chunwei (all five inventors)
  • Assignee: AFFYMAX TECHNOLOGIES N.V.
  • Correspondent: [not retrieved] — cannot assess recurrence without the field.
  • Context: Inventor-to-company assignment capturing the invention into the Affymax corporate family, executed ~9.5 months after filing.

• 1996-08-26 / recorded [not retrieved] — Reel [not retrieved]

  • Conveyance: Assignment
  • Assignor: AFFYMAX TECHNOLOGIES N.V.
  • Assignee: AFFYMETRIX, INC.
  • Correspondent: [not retrieved]
  • Context: Intra-family transfer following the Affymax→Affymetrix spin-off; this is the link that gave Affymetrix its chain of title (confirmed in the Delaware opinion: "In 1996, after Affymax had spun off Affymetrix, Affymax assigned its rights in the application for the '716 patent to Affymetrix").

• 1999-02-08 / recorded [not retrieved] — Reel [not retrieved]

  • Conveyance: Merger
  • Assignor: AFFYMETRIX, INC., a California corporation
  • Assignee: AFFYMETRIX, INC., a Delaware corporation
  • Correspondent: [not retrieved]
  • Context: Internal reorg — corporate reincorporation/merger of California Affymetrix into Delaware Affymetrix; no change in beneficial owner.

• 2012-06-27 / recorded [not retrieved] — Reel [not retrieved]

  • Conveyance: Security Agreement
  • Assignor: AFFYMETRIX, INC.
  • Assignee: GENERAL ELECTRIC CAPITAL CORPORATION, AS AGENT
  • Correspondent: [not retrieved]
  • Context: Securitization — collateral grant over the IP portfolio to a lending agent; not a transfer of beneficial ownership. (Note: postdates the July 2004 Illumina suit, so it is irrelevant to the pre-litigation chain.)

• 2015-11-13 / recorded [not retrieved] — Reel [not retrieved]

  • Conveyance: Release by Secured Party
  • Assignor: GENERAL ELECTRIC CAPITAL CORPORATION, AS AGENT
  • Assignee: AFFYMETRIX, INC.
  • Correspondent: [not retrieved]
  • Context: Termination/release of the 2012 security interest; encumbrance cleared.

Continuation filings are not assignments. The record also shows priority relationships to US 5,974,164 (filed 1995-10-16), US 6,600,996, US 6,733,964, US 6,242,180 and US 6,607,887 — these are continuation/divisional family members stemming from 08/327,525, not transfers of the '716 patent itself.

---

Timeline diagram

timeline
    title Ownership of US 5795716
    1994 : Application filed by individual inventors
    1995 : Inventors assign to Affymax Technologies NV
    1996 : Affymax assigns rights to Affymetrix Inc
    1998 : Patent issues to Affymetrix
    1999 : Affymetrix California merges into Affymetrix Delaware
    2004 : Affymetrix sues Illumina for infringement
    2007 : Jury verdict for Affymetrix
    2012 : Security agreement with GE Capital
    2015 : GE Capital releases security interest
         : Patent reaches anticipated expiration
    2016 : Affymetrix acquired by Thermo Fisher

---

NPE / troll-pattern signals

1. Shell-entity transfer — not present. The chain runs Affymax Technologies N.V. → Affymetrix, Inc. → Affymetrix, Inc. (DE). Every post-1996 link terminates at a named operating company with commercial products; there is no "IP Holdings / Licensing / Ventures" entity anywhere in the chain, and no single-purpose LLC. The only non-operating party is General Electric Capital Corporation, as Agent (2012-06-27), which is a secured lender, not an acquirer.

2. Known asserter in the chain — not present. No assignee matches the Acacia / Marathon / IV / IPNav / Wi-LAN / Conversant / Vringo / Pendrell / Round Rock lists. The sole asserting party of record is Affymetrix, Inc., which asserted the '716 in its own name against a direct competitor (Illumina), and which appears on RPX/Unified directories as a defendant-side operating company, not as a high-frequency NPE plaintiff.

3. Repeat correspondent across the chain — unclear (not assessable). The correspondent-of-record field was not retrievable for any of the five events, so I cannot test recurrence. This is a data gap, not a negative finding — I am marking it unclear rather than "not present."

4. Cascading transfers — not present. Five recorded events span 1995-08-04 → 2015-11-13, i.e. ~20 years. There is no burst of chained LLC transfers in a sub-24-month window. The two ownership transfers (1995 and 1996) are 12.7 months apart but are both within the Affymax/Affymetrix corporate family, not shell hops.

5. Pre-litigation transfer — not present. Suit was filed 2004-07-26 (D. Del. C.A. No. 04-901-JJF). The immediately preceding ownership transfer was the 1999-02-08 reincorporation merger — ~5.5 years before filing, and a change of incorporation state rather than a change of beneficial owner. Nothing in the six months before July 2004 touches title; the only 2004-era record is Affymetrix's own corporate security/financing activity, which is unrelated to venue or standing engineering.

6. Bankruptcy fire-sale — not present. Affymetrix never filed Chapter 7 or 11. The end-of-life event is an acquisition by Thermo Fisher Scientific (2016) — a going-concern sale of an operating company, and in any event occurring after the patent's 2015-08-18 anticipated expiration.

7. Privateering — not present. The patent was asserted by its beneficial owner in the owner's own name, against an unaffiliated competitor, with the owner funding the litigation and taking the damages. There is no NPE fronting for Affymetrix.

8. Defensive aggregator — not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. (Illumina did use ex parte reexamination — control nos. 90/008,885; 90/008,889; 90/008,888; 90/008,886 for the '432, '243, '531 and '716 patents respectively, filed 2007-10-19 — but reexamination is a validity attack, not a defensive acquisition.)

---

Verdict

Operating-company assertion.

Affymetrix, Inc. — a public operating company (NASDAQ: AFFX) selling GeneChip arrays, scanners and analysis software that practice these claims — owned US 5,795,716 by an unbroken chain (inventors → Affymax 1995-08-04 → Affymetrix 1996-08-26 → Affymetrix Delaware 1999-02-08) that contains not a single licensing entity, and asserted the patent in its own name against a direct competitor in Affymetrix v. Illumina, D. Del. C.A. No. 04-901-JJF (filed 2004-07-26), winning a 2007 jury verdict covering the '716 among five patents-in-suit. The only assignment activity in the relevant window is a 1999 reincorporation merger and a 2012–2015 lender security agreement and release — neither of which is a transfer to an asserter. The patent expired 2015-08-18, so post-issuance ownership is now moot.

Caveat: the verdict rests on dates, conveyance types and party names only. Reel/frame numbers and correspondents of record for all five events remain unverified in this analysis and should be pulled from Assignment Center before this is relied on for any filing. Two items warrant a manual check there: (a) the correspondent field on the 1995-08-04 and 1996-08-26 entries, and (b) whether any assignment was recorded after the 2015-11-13 release in connection with the 2016 Thermo Fisher acquisition of Affymetrix.

Verify at: https://assignmentcenter.uspto.gov/ (search patent number 5795716) — mirrored at https://assignment.uspto.gov/patent/index.html; family litigation record at https://patents.google.com/patent/US5795716/en and https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A04-cv-00901

I'll search for the USPTO record and the prior-art citations for US 5,795,716 specifically.

Let me get the actual "Patent Citations" list from the USPTO/Google Patents record for 5,795,716 specifically.

Scope note and search provenance

I searched for the specific number 5795716 and excluded look‑alikes. What came back:

• Google Patents https://patents.google.com/patent/US5795716A/en — the record matches the previously generated memo: US 5,795,716 A, application 08/327,525, filed 1994‑10‑21, granted 1998‑08‑18, inventor Mark S. Chee, current assignee Affymetrix Inc.
• Espacenet cited‑documents page for US5795716 https://pt.espacenet.com/publicationDetails/citedDocuments?CC=US&NR=5795716A&KC=A&FT=D&ND=10&date=19980818&DB=EPODOC&locale=pt_PT — header reads "Lista de todas as citações: US5795716 (A) ― 1998‑08‑18 / 15 documentos citados" under "PATENTES CITADAS NO RELATÓRIO DE PESQUISA" (patents cited in the search report).
• OSTI/DOE Patents record 871785 https://www.osti.gov/doepatents/biblio/871785 — lists four "Works referenced in this record" (non‑patent literature).

Disambiguation warning (important). My keyword searches also surfaced a "Patent Citations (10)" table listing US4592089, US4709212, US4909920, US5420628, US5275710, US5470710, US5571643, WO1996041567A2. That table belongs to US 6,535,624 (gel‑electrophoresis image combining), not to '716 — '716 appears on that page only as a downstream citation. Do not attribute those references to 5,795,716.

Honest limitation. Espacenet reports 15 cited patent documents. The search snippet returned items 1–8 only; the remaining 7 were truncated and I could not retrieve them. I will not enumerate or guess unverified entries. I could not open a full USPTO Patent Center / PatFT full‑text page in this session, so the citation list below is from the Espacenet citation record and the OSTI record, not from the printed patent front page.

---

Part A — Examiner/search‑report patent citations retrieved (items 1–8 of 15)

---

Part B — Non‑patent literature cited in the '716 record (OSTI/Google Patents)

All four are printed publications published more than one year before the 1994‑10‑21 filing date, so each is a §102(b) statutory bar reference against every claim, not merely §102(a) art.

---

Part C — Bottom‑line §102 assessment

1. No confirmed reference anticipates claims 1–10 as a whole. The retrieved patent citations split cleanly into two groups: (a) biological‑image‑analysis / pattern‑classification software (Bacus US 4,741,043; Ruitenberg US 4,965,725; Stockham US 5,273,632) and (b) array chemistry and sequencing‑by‑hybridization assay (Southern WO 89/10977; Macevicz US 5,002,867; Fodor WO 92/10588; Drmanac US 5,202,231; Winkler WO 92/10092). Neither group discloses the claimed limitation that is the heart of '716: code that compares probe intensities to each other (or a higher intensity to a lower intensity) and generates a base call from that comparison — the intensity‑ratio, reference, and statistical engines described in §II–§IV of the specification.
2. Claims 2 and 6 (the "ratio" families) are the hardest to anticipate: the recited "ratio of a higher probe intensity to a lower probe intensity … compared to a predetermined ratio value" is a numerical‑comparison limitation that none of the retrieved references is titled or described as disclosing. They would require full‑text verification before any §102 assertion.
3. Claims 3 and 7 (reference‑method) are the ones where the Southern 1992 Genomics paper and WO 89/10977 are the only plausible §102(b) candidates, because their subject matter is comparing hybridization of different sequences. Still unverified in full text.
4. Claims 1, 4, 5, 8 and dependent claims 9–10 (array limitation; fluorescent‑intensity limitation) map onto the array patents (WO 92/10588, US 5,202,231, WO 92/10092) for the probe‑array and fluorescence elements, but those references do not supply the recited comparison/generation code. Expect these to be §103 references, not §102.
5. Own‑work overlap. WO 92/10588 and WO 92/10092 name Affymax as applicant — the same entity to which '716 was assigned on 1995‑08‑04. Under pre‑AIA §103(c), commonly owned §102(e)/§102(f)/§103(a) art is disqualified for obviousness purposes; it remains available under §102(a)/(b) as publications. This is a real tactical point if these references were asserted.
6. Not prior art. US 5,974,164; EP 0 717 113 A2; US 6,600,996; US 6,733,964; US 6,242,180; US 6,607,887 and publications US 2003/0220748 A1 and US 2005/0186581 A1 are later family members/continuations of the '716 application (per the Google Patents family timeline in the prior memo) and cannot be §102 art against it.

---

Part D — Explicit uncertainties

• Items 9–15 of the 15 search‑report citations were not retrieved. The Espacenet snippet cut off after item 8. I decline to fill this gap by inference. To complete the list you need the printed front page of US 5,795,716 (or the Patent Center "References Cited" table), which I could not open in this session.
• Item 8's patent number is unverified; the "WO 92/10092" identification is my inference from the title matching a document the specification itself cites.
• US 4,741,043(B1), 1994‑08‑09 is reported literally as Espacenet shows it. A "B1" on a 1988‑granted patent at that date is unusual (possibly a reexamination certificate); I cannot confirm what it is.
• I have full text for none of the cited references. The §102 column above is a subject‑matter assessment from titles, dates and known field content, not a full‑text claim chart. Any §102 assertion — especially against claims 2/6 and 3/7 — must be re‑run against the reference full texts.
• The Delaware 1:04‑cv‑00901 litigation flag and the Affymetrix v. Illumina "'716" discussion noted in the prior memo remain unverified as to this patent; nothing in this search changed that.

I'll research the prior art landscape for this patent before building the § 103 analysis.

Let me pull the record's cited prior art and the key contemporaneous references.

Obviousness Analysis — US 5,795,716 (pre‑AIA 35 U.S.C. § 103)

0. Note on the "Prior Art section" — and what I actually used

The page text supplied to me does not contain a "Prior Art" section with a reference list. What it contains is the machine-generated "Prior art keywords" line (base, nucleic acid, probe, sequence, intensity) plus the legal-events/litigation timeline. That is not a prior-art citation list.

Because the instruction was to use the record's prior art, I retrieved the search-report citation list for US 5,795,716 — the 15 documents cited in relation to the patent — from Espacenet:

https://pt.espacenet.com/publicationDetails/citedDocuments?CC=US&NR=[5795716A](/patent/5795716A)&KC=A&FT=D&ND=10&date=19980818&DB=EPODOC&locale=pt_PT
("Lista de todas as citações: US5795716 (A) ― 1998-08-18 — 15 documentos citados")

I also used the non-patent literature listed on the patent's OSTI bibliographic record (https://www.osti.gov/doepatents/biblio/[871785](/patent/871785)) and the references the '716 specification cites in its own text (US 5,143,854; PCT WO 92/10588; PCT WO 92/10092; application Ser. No. 08/249,188).

Critical caveat on verification: I verified the titles, inventors, assignees, priority dates and publication dates of the 15 cited documents from the Espacenet list. I did not read the full text of each reference in this session. My characterizations of their disclosures are therefore based on title/abstract/field knowledge and the patent's own characterizations of them, and are marked as such. Before filing anything, pull the full texts. Where I am uncertain, I say so.

Note also: the earlier-generated summary of this patent states the full claim set (claims 11 and higher) was never verified. This obviousness memo inherits that limitation — I analyze claims 1–10 as previously reconstructed, and flag the unverified region in §4.6.

---

1. Governing legal standard

• Pre-AIA § 103 applies. Application 08/327,525 was filed 21 Oct 1994 and the patent issued 18 Aug 1998; the AIA first-inventor-to-file § 102/103 regime does not govern. Pre-AIA § 103(c) (the version in force, covering only § 102(f)/(g) art) means common ownership with the Affymax/Affymetrix family does not disqualify § 102(a)/(b)/(e) art — so the same-family Fodor/Pirrung/Chee references remain available, subject to § 102(e) timing analysis.
• Graham v. John Deere Co., 383 U.S. 1 (1966): scope and content of the prior art, differences between the claims and the art, level of ordinary skill, and secondary considerations.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): the motivation to combine need not be taught in the references; it can be a "design incentive," "market pressure," "common sense," or the recognition that "a technique … used to improve one device" would "improve similar devices in the same way." A combination of known elements is obvious where it "does no more than yield predictable results." Teaching-suggestion-motivation remains one permissible rationale (per KSR / In re Kahn), not the only one.
• Level of ordinary skill (PHOSITA, 1994): in this art the "person" is realistically a team — a molecular biologist/biochemist experienced in hybridization-based sequence analysis plus a software engineer capable of image/signal processing. The '716's recorded assignment names exactly such a mixed team (Chee, Wang, Jevons, Bernhart, Lipshutz). This matters: it makes cross-disciplinary combination (hybridization chemistry + computer image analysis) more likely to be obvious, not less.

---

2. The claimed subject matter in § 103 terms (building on the prior summary, not repeating it)

From the previously generated section, the ten claims divide into four analytic families, each expressed twice:

Two breadth observations drive everything below:

1. Claim 1 does not require an array and does not require fluorescence. Claim 9 adds the array limitation; claim 10 adds fluorescence. So claim 1 is a bare "read intensities → compare → call base" software claim. It is the most vulnerable.
2. Claims 2–4 add algorithm steps (ratio vs. cutoff; cross-set comparison; comparison to statistics). These are the harder claims, but each added step is a known data-analysis technique applied to a known data set.

---

3. Prior art of record (15 cited documents) and prior-art status as of 21 Oct 1994

Status key: §102(b) = patent/printed publication more than one year before filing; §102(a) = before the invention; §102(e) = US patent granted on an earlier-filed US application (or a pre-AIPA § 102(e) equivalent). "Not art" = postdates the 21 Oct 1994 filing and cannot be § 102(a)/(b); pre-AIPA § 102(e) did not reach published applications, so published PCT applications postdating the filing drop out. This timing analysis is mine, based on the dates in the Espacenet list.

Non-patent literature of record (OSTI record for US 5,795,716):

• Drmanac, Labat & Crkvenjakov, "An Algorithm for the DNA Sequence Generation from k‑Tuple Word Contents of the Minimal Number of Random Fragments," J. Biomol. Struct. Dyn. 8(5) (Apr 1991).
• Strezoska, Paunesku, Radosavljevic et al., "DNA sequencing by hybridization: 100 bases read by a non-gel-based method," PNAS 88(22) (Nov 1991).
• Drmanac, Drmanac, Strezoska et al., "DNA sequence determination by hybridization: a strategy for efficient large-scale sequencing," Science 260(5114) (Jun 1993).
• Southern, Maskos & Elder, "Analyzing and comparing nucleic acid sequences by hybridization to arrays of oligonucleotides: Evaluation using experimental models," Genomics 13(4):1008–1017 (Aug 1992) — note the title: analysing and comparing.

All four are § 102(b) printed publications.

---

4. Element-by-element analysis and the combinations

4.1 Claims 1 and 5 — "compare the intensities to each other, then call the base"

Primary combination for claim 1: WO 92/10588 (Fodor/Solas) + US 5,202,231 (Drmanac), optionally + US 4,741,043 (Bacus) or US 4,965,725 (Rutenberg).

• WO 92/10588 supplies the probe set, the hybridization, the addressable known probe sequences and the raw signal data.
• Drmanac supplies the analysis step that converts a set of hybridization signals into a sequence determination, in the very same SBH field.
• Bacus/Rutenberg supply the computerized intensity comparison/classification step that claim 1 recites ("comparing said plurality of probe intensities to each other"). Their citation in the search report is itself evidence that the examiner regarded computer image/intensity analysis as within the pertinent art.

Why combine (KSR rationale): WO 92/10588's arrays produce hundreds-to-millions of intensity data points that cannot be read by eye; automating the read-out is not merely suggested, it is necessary to use the reference at all. The image-analysis art (Bacus/Rutenberg) existed precisely to automate classification of biological-specimen intensity data. A POSITA automating an array read-out would have looked to it as a matter of ordinary design.

4.2 Claims 2 and 6 — ratio of higher to lower intensity vs. a predetermined cutoff

Combination: WO 92/10588 + US 5,202,231 + US 5,273,632 (Stockham), or + US 4,741,043 / US 4,965,725.

• US 5,273,632 (Stockham) is the cleanest fit: it is a computer method for analysing a test signal profile against a reference, quantifying and thresholding signal values. That is the "compare a higher value to a lower value against a cutoff" mechanism in a different signal-processing context.
• The Macevicz US 5,002,867 approach (differential hybridization of probe mixtures) presupposes that the relative strength of binding distinguishes the correct base.
• The Southern Genomics 1992 paper is a § 102(b) publication evaluating hybridization to oligonucleotide arrays "using experimental models," i.e., quantifying how strongly matched vs. mismatched probes signal.

Why combine: The fundamental chemistry of SBH is that a perfect-match duplex melts higher / binds more label than a single-mismatch duplex. Any POSITA reading the array therefore had to set some discrimination threshold between the top signal and the next signal. Choosing to express that as a ratio (a normalization that cancels spot-to-spot brightness variation) and comparing it to a predetermined cutoff is a routine engineering choice — the kind of "predictable variation" KSR holds obvious. The patent's own specification concedes the point by calling the ratio method "very accurate when there is good discrimination" and by describing the cutoff as a number "generally need[ing] to be adjusted to produce optimal results" — i.e., a tuning parameter, not an inventive insight.

4.3 Claims 3 and 7 — compare sample intensities to a reference sequence's intensities

Combination: WO 92/10588 + WO 89/10977 (Southern) + US 5,202,231 (Drmanac), optionally + US 5,273,632.

This is the strongest obviousness case of the four families:

• Southern WO 89/10977 is literally directed to "Analysing polynucleotide sequences" by hybridization, and the Southern Genomics 1992 paper is titled "Analyzing and comparing nucleic acid sequences by hybridization to arrays of oligonucleotides."
• Drmanac US 5,202,231 / 5,492,806 / 5,525,464 concern comparing hybridization results across probe sets to determine sequence.
• Stockham US 5,273,632 supplies the general signal-processing rationale of comparing a test profile to a reference profile.

Why combine: Comparing an unknown/test sample against a known reference is the canonical control paradigm of molecular biology (Southern blots, sequencing ladders, allele-specific controls) and of clinical diagnostics. A POSITA facing the known problem of cross-hybridization — which the art itself acknowledged, e.g., the family's EP 0 705 271 describes "substantial difficulty in analyzing probe arrays hybridized to a target nucleic acid due to the hybridization of partially mismatched sequences" — would predictably use a known reference sequence as the comparison baseline. Claim 3 recites no more than that comparison.

4.4 Claims 4 and 8 — compare sample intensities to statistics across multiple experiments

Combination: WO 92/10588 + US 4,965,725 (Rutenberg) + US 4,741,043 (Bacus) + US 5,273,632 (Stockham).

• Rutenberg is automated statistical/neural classification built from a body of training data — i.e., deriving statistical decision criteria from many prior observations.
• Bacus is quantitative image-feature analysis.
• Stockham compares test data to a reference/standard.
• The Drmanac/Strezoska/Southern publications of record establish that quantitative, reproducibility-focused analysis of hybridization data was an active field before 1994. (Also on point in the same field: Elder, "Analysis of DNA oligonucleotide hybridization data by maximum entropy" (1992), and Lipshutz, "Likelihood DNA sequencing by hybridization" (1993), both § 102(b) publications.)

Why combine: Fluorescence intensity varies run-to-run with labelling efficiency, hybridization stringency, scanner gain and photobleaching. The obvious remedy for measurement variability is replication plus statistics (mean and standard deviation of controls). Claim 4 recites exactly that. Under KSR, applying a known statistical technique to a known measurement to improve reliability is the paradigm of predictable variation; "obvious to try" is satisfied because the number of candidate remedies (threshold on raw signal, threshold on ratio, compare to a single reference, compare to the distribution of reference runs) is small, finite, and predictable.

4.5 Claims 9 and 10 — array of probes; fluorescent intensities

• Claim 9 (array): WO 92/10588 (matrix of defined oligos); US 5,143,854 (arrays of ≥100 different polymers); US 5,445,934 (oligonucleotide arrays); US 5,527,681; US 5,384,261.
• Claim 10 (fluorescence): US 5,143,854 expressly claims a receptor "comprising a marker selected from radioactive markers and fluorescent markers"; WO 92/10588 and US 5,470,710 (fluorescent multiplex DNA sequencing) also.

These are pure structural add-ons; obviousness of 9 and 10 follows a fortiori from 1–8.

4.6 If unverified claims cover "pooling" or the comparative-visualization GUI

The earlier summary could not confirm claims above 10. If the patent does contain claims to (a) dual-label simultaneous hybridization of reference + sample, or (b) the two-pane reference/sample display:

• Pooling/two-colour: US 5,800,992 (Fodor et al., "Method of detecting nucleic acids," priority 1989-06-07, issued 1998-09-01) is § 102(e) art and is in the same family; WO 92/10588 also contemplates labelled-target detection generally. But flag: US 5,800,992 is commonly owned with the '716, and common ownership raises a § 103(c) argument only for § 102(f)/(g) art — it does not disqualify § 102(e) art, so it stays available. Verify the pooling claims actually exist before relying on this.
• GUI: Bacus and Rutenberg both concern computer display/analysis of specimen imagery; a two-pane sequence-comparison display, absent specific claim detail, would be vulnerable to routine-UI-design arguments. This is the weakest-grounded part of the analysis because the claims are unverified.

---

5. The motivation to combine, stated in KSR form

A rejection can be articulated on several independent rationales, any one of which suffices:

1. Same field of endeavor / same problem. WO 92/10588, the Drmanac references, Southern WO 89/10977, Macevicz and the '716 all address the identical problem: reading a base from hybridization to defined probes. Combination within the field is presumptively suggested.
2. The reference requires the combination to work. WO 92/10588 generates far more intensity data than can be interpreted manually; a POSITA must automate.
3. Predictable variation of a known technique. Ratio thresholds, reference comparison and statistical normalization are known data-analysis tools; applying them to array intensity data yields no more than the predictable result (fewer miscalls, better discrimination).
4. Recognized, documented problem in the art. The literature of record acknowledges the mismatch/cross-hybridization problem; the claimed methods are predictable attempts to overcome it.
5. Design incentive / market pressure. The specification itself frames the driver as clinical/genotyping use (HIV protease, AZT resistance, four HIV-1 clones, pre-/post-ddI patient samples). A diagnostic market demands reproducible calls — a direct incentive to normalize against references and replicate runs.

---

6. Where obviousness is weakest — the patent owner's best counterarguments

Intellectual honesty requires flagging these. A well-briefed patent owner would argue:

1. Claim 1 is broad, but the specific algorithmic content of claims 3–4 and the second reference-method implementation is not in the cited art as I can presently verify. In particular:
   • the ratio-of-ratios construction (per-base ratio of sample ratio to reference ratio),
   • the normalized-difference approach with neighboring-position analysis (FIG. 5C/5D, the "characteristic pattern … of a mutation"),
   • per-base standard-deviation cutoffs in the statistical method,
   • the cascading fall-through to IUPAC ambiguity codes.
     If claims are drafted to these specifics (or if the only verified claims are the broad ones and later claims add them), the obviousness case must be re-run claim-by-claim against full reference texts.
2. Unexpected results. The specification's own examples assert that the reference method correctly called a mutant base that the intensity-ratio method miscalled (FIG. 12) and that the intensity-ratio method correctly tracked a mutation across a 0→100% mutant titration (FIG. 11). If the patent owner can tie those results to a claimed limitation with adequate nexus, that is a Graham secondary-consideration argument. I cannot evaluate the strength of that nexus from the record I have.
3. § 102(e)/§ 102(b) timing attacks on the record's own art. Four of the 15 cited documents (WO 95/11995, WO 95/35505, and — depending on how one reads the pre-AIPA § 102(e) provision — the two post-filing-published family items) postdate the 21 Oct 1994 filing and are not prior art. A challenger (or the patentee) must prune the list. This cuts in the patentee's favour to the extent it removes supporting references, and against the patentee to the extent it removes its own "background" admissions.
4. Common ownership. Several of the array references (US 5,445,934; US 5,527,681; US 5,384,261; WO 92/10588) are Affymax/Affymetrix-family art. As noted, pre-AIA § 103(c) disqualified only § 102(f)/(g) art, so they remain available — but a litigator preferring clean third-party art will lean on Drmanac, Southern, Macevicz, Bacus, Rutenberg, Stockham and Weiss/Kimball, which are independent of the patentee. My combinations above are built to be resilient on that point.

---

7. Bottom line

The single most powerful attack, if forced to pick one, is WO 92/10588 (Fodor/Solas) in view of Southern WO 89/10977 and/or the Drmanac SBH family, further in view of the computer image-analysis art (Bacus, Rutenberg), on claims 1 and 5 — because claim 1 is a bare "receive intensities → compare → call base" software claim with no array and no fluorescence limitation.

---

8. Explicit uncertainties (do not over-read this memo)

1. I did not obtain the "Prior Art section" the prompt referred to. The supplied page text contained only prior-art keywords, not a reference list. I substituted the Espacenet search-report citation list (15 documents) and the OSTI NPL record. If a different "Prior Art" list was intended (e.g., a Google Patents "Citations/References Cited" block, or prosecution-file art not on the face), my combination analysis may need to be re-based on it.
2. I did not read the full texts of the cited references in this session. Disclosures are characterized from titles, abstracts and field knowledge. Full-text verification is required before any rejection is advanced on the record.
3. The claim set above claim 10 is unverified (inherited from the prior summary). The 43-claim figure I encountered belongs to continuation US 6,733,964, not to the '716. Do not attribute '964's claims to the '716.
4. The § 102(e) timing conclusions in §3 are my analysis, not a citation to an authority; a pre-AIA § 102(e) / § 103(c) question should be confirmed with the applicable version of the statute in force in 1994.
5. No reexamination, reissue, IPR or 2026 Federal Circuit activity for US 5,795,716 was found in my searches, and I therefore express no opinion on whether any claim of the '716 has ever been cancelled or confirmed in an adversarial proceeding. The Delaware family litigation flag (1:04‑cv‑00901) remains an unverified lead as to whether this patent was asserted.