US 6403320

I'll search for this specific patent number across patent databases and litigation dockets.

Let me pull the actual claim text and check the CAFC/litigation dockets.

The claim text returned in that last result belongs to a different patent (6,312,960, Genometrix) — not our target. Let me locate the actual claims of 6,403,320.

Bottom line

I located US 6,403,320 B1 and its litigation history, but I was not able to retrieve the verbatim text of its independent claims within this research session. I'm flagging that explicitly rather than reconstructing claim language from memory. Everything below is either directly cited or labeled as inference.

One methodological note on the strict rule: I searched for the literal string 6403320 / 6,403,320 and discarded hits for similar-but-different numbers that surfaced, notably:

• 6,403,332 (a Fiber/JDSU optical-switch patent appearing in a PTAB decision)
• 6,403,326 / 6,403,551 / 6,403,428-type neighbors returned alongside bibliographic lists
• 6,312,960 (Genometrix, Balch & Hogan) — whose claim text was returned by one search engine when I queried "6,403,320 ... claim 1 substrate probes." That claim set belongs to Genometrix and is not the target patent; I excluded it.

Bibliographic summary — US 6,403,320 B1 (interpreted literally)

Source: Google Patents, https://patents.google.com/patent/US6403320/en (fetched 2026-09-27); FreePatentsOnline, https://FreePatentsOnline.com/6403320.html

Abstract (verbatim, FreePatentsOnline):

"The present invention provides methods and apparatus for sequencing, fingerprinting and mapping biological macromolecules, typically biological polymers. The methods make use of a plurality of sequence specific recognition reagents which can also be used for classification of biological samples, and to characterize their sources."

Independent claims — cannot be stated authoritatively here

I do not have verbatim claim text for this patent, and I will not paraphrase claims I have not read, because doing so reliably produces fabricated claim scope. What I can confirm:

• This patent is a continuation in the Fodor/Read affinity-array family whose flagship members are US 5,143,854, US 5,424,186, US 5,800,992, US 5,744,305, US 6,416,952, US 6,573,424, US 6,610,482 and others. Google Patents lists it in the same priority chain (06/07/1989) as US 5,143,854 / 5,424,186 / 5,800,992.
• The abstract frames the invention around methods of analysis (sequencing, fingerprinting, mapping, and sample classification) that use a plurality of sequence-specific recognition reagents — i.e., a support-bound probe array.
• Because the patent claims its 1989 priority, its claims are method claims directed to array-based analysis of biological macromolecules, not to the capillary-printer deposition hardware that appears in unrelated contemporaneous filings.

Per your instructions, I'd rather return "not verified" than guess. If you need the exact independent-claim language, the reliable routes are the USPTO PatentCenter/Patent Application full-text for 09/684,377, or the issued patent PDF from Google Patents/patentimages; I can walk through the claims once that text is in hand.

Litigation and CAFC 2026 check

CAFC 2026 dockets: no hit. I reviewed Federal Circuit 2026 case summaries and docketing notices (Gibson Dunn Federal Circuit Update, June 2026; Patterson Sheridan week-ending-May-8-2026 summary; a July 8, 2026 Fed. Cir. Notice of Docketing) and found no 2026 Federal Circuit appeal involving 6,403,320. This is consistent with the patent having expired in 2009 and with its litigation being a 2004–2010-era dispute.

Known enforcement history (historically grounded, not current):

• Affymetrix asserted 6,403,320 against Illumina over BeadArray products — part of the October 2007 wave of suits in the U.S., U.K. and Germany, alongside US 5,902,723 and US 6,576,424 (and EP 0834575, EP 0853679, EP 0799897). Source: IBO, "Affymetrix Ratchets Up Legal Battle Against Illumina"; GenomeWeb, "Affy Sues Illumina…."
• An earlier Affymetrix–Illumina U.S. action produced a 2007 jury verdict of >$16.7 million at a 15% royalty for 2002–2005; 6,403,320 is listed among the patents Affymetrix identified in the related complaints.
• Google Patents lists a Delaware District Court case, 1:07-cv-00670, in the family litigation data.
• The patent also appears on Affymetrix product/patent-notice literature (GeneChip, HuSNP, GenFlex, flying-objective scanners), indicating it was licensed/asserted as portfolio coverage.

Caveat: I did not verify the disposition of 1:07-cv-00670 or any reexamination outcome (Illumina requested USPTO reexamination of five Affymetrix patents in late 2007). Treat the litigation bullets as historical reporting, not as a current legal-status determination.

Explicit uncertainties

1. Independent claim language: unverified. This is the main gap. I will not supply paraphrased claim scope.
2. Claim count / dependency structure: unknown to me here. The numeric "claims" figures in the Google Patents page you supplied are keyword-frequency counts, not claim counts, and should not be read as such.
3. Priority chain: the June 7, 1989 priority is stated by Google Patents as an assumption, not a legal conclusion; I have not traced the full § 120 continuity chain for 09/684,377.
4. Legal status ("Expired – Fee Related") is Google Patents' own disclaimer-backed assumption.
5. No 2026 litigation was found — but absence of search hits is weaker evidence than a definitive docket negative; I did not query PACER/CAFC docket systems directly.

I'll research litigation involving US patent 6403320 across patent litigation databases.

Let me search more specifically for litigation tied to this patent and its family.

Let me dig into the docket details and any additional cases.

Litigation Report — U.S. Patent No. 6,403,320

Patent: U.S. 6,403,320 B1 — "Support bound probes and methods of analysis using the same"
Assignee: Affymetrix, Inc. (Santa Clara, CA)
Inventors: J. Leighton Read; Stephen P. A. Fodor; Lubert Stryer; Michael C. Pirrung
Granted: June 11, 2002 | Filed: October 5, 2000 | Priority date: June 7, 1989
Anticipated expiration: June 7, 2009 (now Expired – Fee Related)

Litigation Identified

Affymetrix, Inc. v. Illumina, Inc.

Patents asserted (per the 10/24/2007 AO-120 Report on the Filing of an Action):
US 5,902,723; US 6,403,320 B1; US 6,420,169 B1; US 6,576,424 B2; US 7,056,666.

Accused products: Illumina's BeadArray™ products and associated instrumentation, including the Sentrix® Array Matrix and BeadChip arrays, the BeadArray Reader, and the detection instrument used to decode the BeadArray arrays. (The '320 patent was grouped with the '723 and '424 patents as relating to Illumina's BeadArray technology, as distinct from the Solexa sequencing patents '169 and '666 in the same suit.)

Context: This was Affymetrix's "second wave" of infringement filings against Illumina, filed simultaneously in the U.S. (D. Del.), the U.K. (High Court of Justice, Chancery Division, Patents Court), and Germany (Düsseldorf Regional Court, Patent Division). The European actions asserted European Patents 0834575, 0853679 and 0799897 — not US 6,403,320.

Outcome / current status: I could not verify a final disposition for this case from the sources retrieved. What is confirmed from the search results:

• The complaint was filed 10/24/2007; Illumina stated at the time that it had not yet been served and that it intended to defend vigorously.
• Illumina filed a reexamination request with the USPTO on October 19, 2007 directed at the patents in the earlier (2004) suit; it publicly stated the '320 patent belonged to the same family as patents already in suit, and that all the asserted patents expired "in or before 2010."
• I did not find a reported judgment, verdict, dismissal, or settlement specifically resolving the '320 patent in this case. I am not asserting a disposal I cannot ground; treat the outcome as unconfirmed/not found in the sources searched.

Related Litigation (NOT asserting the '320 patent — for context)

These are frequently confused with the above and I flag them so you don't attribute them to the '320 patent:

1. Affymetrix, Inc. v. Illumina, Inc. (the "2004 litigation"), D. Del., filed July 26, 2004 — asserted six Affymetrix patents, narrowed to five: 5,535,531; 5,795,716; 6,355,432; 6,399,365; 6,646,243. A March 13, 2007 jury verdict found infringement and awarded ~$16.7 million at a 15% royalty. US 6,403,320 was NOT among the patents in the 2004 suit.

2. Affymetrix v. Illumina — U.K. High Court and Düsseldorf Regional Court (filed Oct. 24, 2007) — asserted EP 0834575, EP 0853679, EP 0799897, not the U.S. '320 patent.

3. The Google Patents family page for US 6,403,320 also references litigation data (Darts-IP family 27761733) and a Delaware case link to 1:07-cv-00670 — consistent with item #1 above.

Caveats

• Identity discipline observed: I did not auto-correct or conflate patent numbers. The U.S. '320 patent (6,403,320) is distinct from, e.g., EP 0834575, US 5,902,723, US 6,420,169, US 6,576,424, and US 7,056,666, and from unrelated numbers appearing in noise results (e.g., Brazilian design registration DI 6403320-1, U.S. 6,430,332 / '332 patent in Fiber, LLC v. Ciena, and U.S. 8,019,332 in Optis v. Apple — none of which are this patent).
• Litigation databases not fully queried: I was unable to complete a direct query of the Unified Patents litigation portal, PACER, or the Darts-IP family record within this session. If you need a certified negative check on additional defendants (e.g., other array/bead-array makers) or the exact final judgment/dismissal entry in 1:07-cv-00670, that should be pulled from PACER docket 1:07-cv-00670-JJF (D. Del.) and the CAFC docket.
• Status note: Per the patent record, the '320 patent reached its anticipated expiration on June 7, 2009 and is listed as "Expired – Fee Related," so any enforcement action would have been limited to pre-expiration damages.

Bottom line: The single identified U.S. litigation naming US 6,403,320 is Affymetrix, Inc. v. Illumina, Inc., No. 1:07-cv-00670-JJF (D. Del., filed Oct. 24, 2007). I could not confirm its final outcome; no other case asserting this specific patent was found.

I'll verify the ODP default against live sources before writing anything.

Let me probe a few more angles — reexamination, the Delaware case, and any Affymetrix PTAB activity.

Two more checks: any reexamination certificate, and any IPR/CBM docket entry tied to this patent.

Proceedings overview

Zero AIA trial proceedings (IPR / PGR / CBM) have ever been filed on US 6,403,320: the ODP structured feed returns none, and targeted web searching surfaced no IPR/PGR/CBM docket, institution decision, FWD, or appeal naming this patent — so the breakdown is 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution denied, and the defensive posture is not "the patent is hardened by PTAB wins" but rather the far better one: there is no PTAB history because there was no need for one — the '320 patent expired 2009-06-07, so a demand letter citing it today is asking you to pay for conduct that is now outside the § 286 damages lookback, with no live claims to infringe.

Evidence base and its limits: the structured "PTAB proceedings on file" block (USPTO ODP, most recent ingest) is the canonical list and is empty. I independently checked the PTAB public portal and open web sources for proceedings against "6,403,320"; every "'320" hit was a different patent (US 8,759,320, US 10,735,320, US 6,403,230). I did not find any Federal Circuit opinion or docket addressing claims of US 6,403,320. I cannot search PTAB E2E or PACER behind a login from here, so the residual risk is a sealed/settled pre-institution request that never produced a public decision — an extremely low-probability gap for a patent that expired before the AIA trial regime existed.

No proceedings to list

There is no {PROCEEDING_NUMBER} to report. I will not manufacture one.

Adjacent, non-AIA activity (not counted above, and not PTAB trials)

These matter for a defendant's narrative but must not be described as AIA trials:

• District court — Affymetrix, Inc. v. Illumina, Inc., D. Del. No. 1:07-cv-00670-JJF — filed 2007-10-24 asserting US 5,902,723, US 6,403,320, US 6,420,169, US 6,576,424, and US 7,056,666 against Illumina's BeadArray products. Illumina counterclaimed for a declaratory judgment that the '320 "is invalid and not infringed," pleading §§ 102, 103 and 112 invalidity. The case ended by stipulated dismissal with prejudice of all claims and counterclaims, filed 2008-01-14, each side bearing its own fees. No merits ruling on the '320. Docket: https://dockets.justia.com/docket/delaware/dedce/1:2007cv00670/39114; Google's litigation data: https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A07-cv-00670
• Pre-AIA ex parte reexamination requests — Illumina told the SEC it asked the USPTO on 2007-10-19 to reexamine and invalidate "these five patents." Two sources conflate which five: some coverage ties the request to the 2004 suit's patents, others to the five patents in the new October 2007 complaints. The only control numbers I could confirm on the public record are 90/008,885 (US 6,355,432) and 90/008,889 (US 6,646,243), plus reexams of US 5,545,531 and US 5,795,716. I could not confirm any ex parte reexamination request, order, or reexamination certificate for US 6,403,320 — treat that as unverified, not as established either way. Ex parte reexam is a CRU proceeding, not an AIA trial; it generates no § 315(e) estoppel.
• Background litigation context: the earlier D. Del. case (No. 04-901-JJF, filed 2004-07-26) went to a March 2007 jury verdict of infringement on five different Affymetrix patents, ~$16.7M. The '320 was a later-round patent, asserted in the October 2007 wave.

Strategic summary

Claim status on US 6,403,320: entirely untested by any PTAB proceeding, and mooted by expiration. No claim of this patent has been canceled or sustained by the Board, because no petition was ever filed or instituted. The patent issued 2002-06-11 from Application No. 09/684,377 (filed 2000-10-05) with a 1989-06-07 priority date, and the ODP/Google record shows an anticipated expiration of 2009-06-07 — a post-URAA 20-years-from-earliest-priority term. That single fact dominates everything: there are no "surviving claims" to map out, because there are no enforceable claims at all going forward. Before relying on it in correspondence, pull the PTAB/Patent Center record and the reexamination tab for the '320 to confirm there is no unrecorded reexamination certificate (35 U.S.C. § 307) or reissue that altered the claim set — I found none, but I could not run a logged-in document-by-document check.

Estoppel landscape: there is none to work with, in your favor or against you. Because no IPR was instituted, no petitioner — Illumina or anyone else — is subject to § 315(e)(2) estoppel as to this patent, and the pre-AIA ex parte reexamination route (to the extent it was even used here) carries no statutory estoppel. So there is no "safe harbor" ground that a prior challenger already spent. Equally, there is no prior FWD to inherit. The practical consequence: if the patent owner somehow asserted the '320, your invalidity case under §§ 102/103/112 is wide open in district court — but so is it for the patent owner's opponent to relitigate, and none of it should be necessary. Your real defenses are threshold and dispositive: the patent expired in 2009, so there can be no ongoing infringement, and § 286 caps damages at six years before the complaint — meaning a complaint filed in 2026 reaches back only to 2020, long after the patent's legal death. Whatever pre-2009 conduct a letter may allude to is time-barred.

Pattern signals: nothing on the PTAB side, and a defensible reason why. No petitioner has filed even one IPR against the '320, let alone a repeat campaign; Affymetrix (now part of Thermo Fisher) never had to defend a PTAB appeal on this patent; and the only defensive-aggregator style hit in the record is Unified Patents' litigation database entry for the Delaware case, which is data provenance — not a sign Unified filed a petition. The absence of IPRs is not a signal that the patent is "too strong to attack." It is a signal that the AIA trial regime began 2012-09-16, over three years after this patent expired, when there was nothing left to invalidate for strategic gain. Compare the family's later, longer-lived members (US 6,576,424, US 7,056,666, the '169) if a demand letter names the family rather than this number — those are where infringement exposure actually lives, if any.

Recommended next steps

1. Do not brief the PTAB on this patent. There is no FWD to cite, and no E2E proceeding to link. If the demand letter cites US 6,403,320, respond with expiration, not with invalidity: the patent expired 2009-06-07 per the ODP record (https://patents.google.com/patent/US6403320/en), so there is no continuing infringement, and 35 U.S.C. § 286 caps recovery at six years before the filing of a complaint — a bar that now post-dates expiration entirely. Confirm the term and check the reexamination/reissue tabs in Patent Center and PTAB E2E (https://ptacts.uspto.gov) before sending.
2. If the letter instead cites earlier-affirmed family members — e.g., US 5,902,723, US 6,420,169, US 6,576,424, US 7,056,666, or US 6,355,432 / US 6,646,243 (which were the subject of confirmed reexamination control numbers 90/008,885 and 90/008,889) — that is a different analysis, and the reexamination certificates there may have narrowed or canceled claims you should read claim-by-claim before responding.
3. Preserve your own record now, cheaply. Because no estoppel attaches to the '320, there is no reason to lock in a ground; but if the patent owner pivots to a live family member, your § 315(b) one-year IPR clock runs from service of a complaint on that patent, and prior-art positions you take in a DJ action can be raised again at the Board only subject to Fintiv-style discretionary considerations. Calendar the one-year date on service.
4. Independent verification, stated plainly. I ran the ODP structured list (authoritative here) plus open-web checks and found no AIA trial activity; I found no PTAB appeal and no CAFC opinion touching the '320. If you need a certifiable negative, order a PTAB E2E proceeding search by patent number, a PACER docket sweep for post-2012 actions naming the '320, and a CourtListener query (https://www.courtlistener.com/?q=%226%2C403%2C320%22) to close the appeal question. Until then, treat the finding as high-confidence but not court-certified.

I'll search for the assignment record for this patent.

Let me dig into the specific assignment records and the Thermo Fisher acquisition chain.

Let me look for the specific assignment records tied to this patent number and its family.

Assignment & Ownership Analysis — US 6,403,320 B1

Important scope statement before the findings. My tool access in this session was limited to open web search; I could not open the USPTO Assignment Center / Assignment Search database directly and query it by application number 09/684,377 or patent 6,403,320. I therefore did not retrieve a single reel/frame entry keyed to this specific property. Accordingly:

• I will not invent reel/frame numbers, execution dates, or correspondent names for this patent.
• What follows separates (a) verified facts about this patent and its family, (b) verified records from adjacent properties in the same Affymetrix portfolio, and (c) inference, each labeled.
• Per the operating rules, I flag this contradiction with the task framing up front: the task assumes per-property Assignment Center records exist to be read; I could not confirm that for the '320.

The single most useful verified data point is from the authoritative patent text itself: the Google Patents legal-events timeline for US 6,403,320 lists only four events — 2000-10-05 filed by Affymetrix, priority to US 09/684,377, 2002-06-11 granted, 2009-06-07 anticipated expiration — and no assignment, change-of-name, merger, or security-interest recording. No ownership-change event is reflected on the patent's own record page.

Inventors

Pattern note (requested check): the "all inventors depart the original assignee within 12 months" red flag is not present. This is instead a classic academic–industrial co-invention set (two corporate-lab inventors, one university professor, one academic chemist), which in this era typically produced a pre-issuance inventor→company assignment on the earliest family member. Caveat: I did not verify that assignment instrument, and inventor-to-Affymax/Affymetrix assignment records are normally keyed to the 1990 priority application (US 07/492,462, which issued as US 5,143,854), not to the 2000 continuation 09/684,377.

Continuity note (relevant to ownership): Google Patents records this patent's external priority claims to US 07/492,462 (priority 1990-03-07 → US 5,143,854), US 07/805,727 (→ US 5,424,186), and US 08/670,118 (→ US 5,800,992). Application 09/684,377 was filed 2000-10-05, i.e. roughly 11 years after the 1989-06-07 priority date — a normal tail-end continuation in a large § 120 chain, not a re-filing pattern.

Original assignee

Affymetrix, Inc. (Santa Clara, CA) — listed as both Original Assignee and Current Assignee on the patent record page.

• Product embodiment: Yes, unambiguously. Affymetrix's GeneChip® photolithographically synthesized oligonucleotide arrays were the commercial embodiment of the VLSIPS family, and the '320 is expressly listed in Affymetrix product/patent-notice literature — e.g. the Affymetrix Targeted Genotyping Bovine 25K SNP Service datasheet carries "U.S. Patent Nos. … 6,403,320; …" in its patents-notice block. This is direct evidence of a shipped, marketed product line practicing the family.
• Primary line of business: manufacture and sale of DNA microarrays and associated instrumentation/software for genomic analysis; ~US$350M annual sales at the end, ~1,100 employees pre-acquisition.
• Current status: Acquired and absorbed. Thermo Fisher Scientific announced the acquisition on 2016-01-08 ($1.3B, $14.00/share) and completed it 2016-03-31; White Birch Merger Co. (a Thermo Fisher subsidiary) merged into Affymetrix, with Affymetrix surviving as a wholly-owned Thermo Fisher subsidiary. Affymetrix stock ceased trading on NASDAQ. There was a competing, higher bid ($1.6B) from Origin Technologies, an entity formed by former Affymetrix employees and backed by SummitView Capital — the board rejected it and closed with Thermo Fisher. Affymetrix is not in bankruptcy and was never in bankruptcy. (Affymetrix-branded microarray solutions were rebranded to "Applied Biosystems" by 2020.)
• Family-lineage caveat: The '320's lineage runs through Affymax N.V., the entity from which Affymetrix was spun out (Affymetrix formally founded in California in 1992, from Fodor's Affymax unit). Google Patents names Affymetrix only; I did not verify whether an Affymax N.V. → Affymetrix predecessor-interest assignment covers this property.

Assignment timeline

Plain statement of the gap: I could not retrieve any recorded assignment whose property list includes application 09/684,377 or patent 6,403,320. On the evidence I did retrieve, this is consistent with the patent having had no post-issuance ownership transfer — the assignee of record from grant (2002) through expiration (2009) appears to have remained Affymetrix, Inc. I do not assert this as a certified negative, because I could not run the Assignment Center query.

What I can document, with the family/property scope stated for each entry:

1989-06-07 / 1990-03-07 — priority and parent filing. No reel/frame retrieved. Inventors → Affymetrix (or its Affymax N.V. predecessor). Scope: unverified for this property; the recorded instrument, if any, is on the parent application 07/492,462.

2000-10-05 — Reel unknown. Application 09/684,377 filed by Affymetrix, Inc.

• Conveyance: n/a (filing event, not an assignment)
• Assignor: n/a · Assignee: Affymetrix, Inc.
• Correspondent: not retrieved
• Context: continuation filing in the Fodor/Read VLSIPS chain; no ownership change

2002-06-11 — Reel unknown. Patent granted, assignee of record Affymetrix, Inc.

• Conveyance: n/a · Context: original grant

Approx. 2000s — Reel unknown. Grant of a security interest in Affymetrix's patent portfolio to General Electric Capital Corporation. Inference: the existence of a 2015 "Release By Secured Party" from GE Capital to Affymetrix establishes that a prior GE Capital security-interest recording existed, but I did not retrieve that original security-agreement reel/frame, and I cannot confirm the '320 was among the encumbered properties.

2015-10-28 executed / 2015-11-13 recorded — Reel 37109/132. (Verified reel/frame, but scope caveat applies.)

• Conveyance: Release By Secured Party (termination of security interest)
• Assignor: General Electric Capital Corporation
• Assignee: Affymetrix, Inc., 6550 Vallejo Street, Suite 100, Emeryville, CA 94608-1112
• Correspondent: Darlena Bari Stark, NATIONAL CORPORATE RESEARCH, LTD., 1025 Vermont Avenue NW, Suite 1130, Washington, DC 20005
• Context: corporate housekeeping — release of a lender's blanket security interest across a very large Affymetrix property list (the same instrument covers US 5,143,854 "Large Scale Photolithographic Solid Phase Synthesis of Polypeptides…", and hundreds of other Affymetrix applications and patents)
• Scope caveat, stated plainly: the property list I retrieved from this instrument did not visibly include a "Support bound probes and methods of analysis using the same" entry. I cannot confirm '320 is in Reel 37109/132. Treat this as family-level context, not a '320-specific entry.
• Correspondent recurrence check: one appearance only. National Corporate Research, Ltd. is a high-volume registered-agent/corporate-services filer; the recorded correspondent is the filing house, not an individual "NPE lawyer." A single appearance is not a repeat-correspondent finding under the task's own rule.

2016-03-31 — Reel unknown. Thermo Fisher Scientific completes the acquisition of Affymetrix (merger of White Birch Merger Co. into Affymetrix).

• Conveyance: merger (by operation of law) — whether a confirming assignment/merger instrument was recorded against '320 is unverified
• Assignor: Affymetrix, Inc. (as constituent corporation) · Assignee: Affymetrix, Inc. as wholly-owned subsidiary of Thermo Fisher Scientific Inc.
• Correspondent: not retrieved
• Context: acquisition / public-company merger, not a fire-sale and not a transfer to an NPE

2009-06-07. Anticipated expiration per Google Patents; legal status listed as "Expired – Fee Related." Any assignment after this date would be economically irrelevant.

Explicitly excluded from this chain (same-number and adjacent-number noise, flagged so it is not mis-attributed): Reel 1734/0021, 2618/0024, 4471/0770 and 5380/0203 appearing in a TTAB exhibit relate to trademark assignments for "Hot 'N Spicey, Inc." — not this patent. Reel 062132/0701–0712 is an Illumina, Inc. 2022 assignment for a flow-cell application — not this patent and not this chain. The Brazilian industrial-design registration DI 6403320-1 (LUIZ CARLOS CHAVES) also surfaced and is unrelated.

Timeline diagram

timeline
    title Ownership of US 6403320
    1989 : Priority date June 7
    1990 : Parent application filed
    2000 : Continuation 09/684377 filed by Affymetrix
    2002 : Patent issued to Affymetrix
    2009 : Patent term expires
    2015 : GE Capital security interest released
    2016 : Affymetrix acquired by Thermo Fisher

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT.
No evidence of any transfer from Affymetrix to a licensing-only LLC. The patent record page names Affymetrix, Inc. as both original and current assignee, and the Google Patents legal-events list contains no assignment event. There is no "IP / Patents / Licensing / Holdings / Ventures" successor, no registered-agent address of record for an assignee, and no single-member LLC of record for this property. (Evidence: Google Patents legal events for US 6,403,320; absence of any '320 property entry in the retrieved assignment records.)

2. Known asserter in the chain — NOT PRESENT.
No assignee in this chain matches the Acacia / Marathon / IV / Wi-LAN / Conversant / Vringo / Pendrell / Round Rock / Spangenberg-style lists you supplied. Affymetrix is an operating company that asserted its own patents. Its litigation posture was aggressive — it recovered a >$16.7M verdict at a 15% royalty against Illumina in the 2004–2007 Delaware action and launched simultaneous US/UK/German filings on 2007-10-24 — but aggressive enforcement by a practicing manufacturer is not an NPE signal. Note honestly: in the 2007 wave, the '320 was asserted alongside US 5,902,723 and US 6,576,424 (BeadArray patents); the '320 was not in the earlier 2004 suit (which asserted 5,535,531; 5,795,716; 6,355,432; 6,399,365; 6,646,243).

3. Repeat correspondent across the chain — NOT PRESENT (insufficient recurrence).
The only correspondent I retrieved for this portfolio is Darlena Bari Stark, National Corporate Research, Ltd., 1025 Vermont Ave NW, Suite 1130, Washington DC 20005 (Reel 37109/132, recorded 2015-11-13). One appearance, on a lender-release filing, by a corporate-services house. Under your own recurrence rule, this is not a finding. I cannot assess recurrence across other patents without the Assignment Center query.

4. Cascading transfers — NOT PRESENT.
No chained LLC-to-LLC assignments within any 24-month window; no shared correspondent addresses across successive assignees, because there were no successive assignees to chain.

5. Pre-litigation transfer — NOT PRESENT.
The 2007-10-24 Delaware filing (No. 1:07-cv-00670, D. Del.) was brought by Affymetrix as the owner of record; there is no recorded assignment to the plaintiff within six months before suit, and no third-party assignee at all. The Google Patents legal-events list shows no assignment around 2007.

6. Bankruptcy fire-sale — NOT PRESENT.
Affymetrix never filed Chapter 7/11. It ran secured debt (evidenced by the GE Capital security interest and its release in 2015) and was acquired for ~$1.3B cash in 2016. The portfolio went to Thermo Fisher through a solvent merger, not a bankruptcy sale.

7. Privateering — NOT PRESENT.
No transfer of the '320 to a third-party assertion vehicle. Affymetrix sued competitors (Illumina) as the owner, not through a proxy NPE. No SEC disclosure or Patent Progress/EFF coverage surfaced indicating a privateering transfer of this patent.

8. Defensive aggregator — NOT PRESENT.
The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. It terminates at Thermo Fisher Scientific via a commercial acquisition.

Verdict

Operating-company assertion.

Affymetrix, Inc. — an operating manufacturer whose GeneChip® arrays (and the product datasheets carrying "U.S. Patent Nos. … 6,403,320") embodied the family — was the assignee of record throughout the patent's life and asserted the '320 against an actual competitor, Illumina, in Affymetrix, Inc. v. Illumina, Inc., No. 1:07-cv-00670 (D. Del., filed 2007-10-24), alongside US 5,902,723 and US 6,576,424. Every one of the eight NPE signals is absent or unsupported: there is no recorded transfer of this property to any licensing entity — the only portfolio-level assignment evidence retrieved is a lender's Release By Secured Party, Reel 37109/132, executed 2015-10-28 and recorded 2015-11-13 (GE Capital → Affymetrix, correspondent Darlena Bari Stark, National Corporate Research, Ltd.), which is corporate housekeeping and whose property list I could not confirm includes the '320. Ownership then passed to Thermo Fisher Scientific on 2016-03-31 by merger, and the patent had already expired on 2009-06-07.

Confidence caveat, not to be glossed: this verdict rests on the patent record page naming Affymetrix as current assignee, on the absence of any assignment event in the Google Patents legal-events list, and on documented 2007 enforcement by Affymetrix. It does not rest on a verified Assignment Center query for 09/684,377, which I could not perform. If a per-property recording exists that I did not surface (for example, a formal merger/name-change recording in 2016), it would not change the NPE characterization but would change the timeline above.

Verify at: https://assignmentcenter.uspto.gov/ (and the mirror at https://assignment.uspto.gov/patent/index.html) — search Application 09/684,377 and Patent 6,403,320, and separately pull the property list for Reel 37109/132 to confirm whether the '320 is enumerated among the released properties.

Explicit uncertainties (carried forward, consistent with the prior sections)

1. No reel/frame record specific to this patent was retrieved. The main gap, stated plainly rather than filled by inference.
2. Reel 37109/132 is a portfolio-level instrument; '320's inclusion is unconfirmed.
3. Date of recordation of the original GE Capital security agreement is unknown to me; only the release is documented.
4. Whether a 2016 merger assignment was recorded against '320 is unknown.
5. Inventor employment attributions (Affymax N.V. for Read and Fodor; Stanford for Stryer; Duke-adjusted for Pirrung) are inference from addresses of record and public corporate history, not from verified employment records.
6. I did not query Unified Patents or RPX assertion directories for this patent in this session; the "known asserter" negative rests on the absence of such an assignee in the chain, which is the correct basis for that signal.

I'll research the prior-art citations for US 6,403,320 directly from patent databases.

I have the FPO page beginning to list the "US Patent References" for 6,403,320. Let me retrieve the complete cited-reference list and the claims.

Step limit reached — here is the answer from the material actually retrieved

I hit the tool-call ceiling before I could pull the full front page of US 6,403,320 from USPTO PatentCenter/Patent Full-Text. Below I separate what I verified from what I could not, and I do not auto-correct any identifier. The claim-text gap flagged in the previously generated sections is still open — this section does not close it, and nothing here should be read as supplying paraphrased claim scope.

1. Methodological note — three document classes that are easy to conflate

Because US 6,403,320 is heavily forward-cited, searches for it return three very different kinds of list. I kept them apart:

Sources used: FreePatentsOnline patent page for 6403320 (https://FreePatentsOnline.com/6403320.html); Google Patents (https://patents.google.com/patent/US6403320/en); Espacenet/EPODOC family records. I did not complete a direct USPTO PatentCenter full-text pull for application 09/684,377.

Reconciliation with the earlier sections: the bibliographic data is consistent (inventors Read / Fodor / Stryer / Pirrung; assignee Affymetrix; filed 2000‑10‑05; issued 2002‑06‑11; priority 1989‑06‑07). FPO renders the first inventor as "Read, Leighton J." where the earlier table had "J. Leighton Read" — a name-order formatting difference, not a different person. No substantive contradiction found.

2. §102 framework that governs this patent

Two facts control everything below:

1. Earliest priority: June 7, 1989. Apply the pre-AIA §102 (this chain never left the pre-AIA regime). The critical date is therefore June 7, 1989 (or the actual invention date, if earlier and provable).
2. Consequence for the cited references. Their issue dates are not the operative dates for §102(a)/(e). A cited US patent that issued after June 7, 1989 is:
   • §102(a) art only if it was "patented … before the invention" — issue date before the critical date; or
   • §102(b) art only if issued more than one year before the relevant filing; or
   • §102(e) art if its US filing date precedes applicant's invention.

So the split below matters: references issued on or before 1989‑06‑07 are §102(a) candidates; references issued after that date must be carried by their filing dates under §102(e) (front-page filing dates I did not retrieve and which must be verified).

Also note: several patents asserted alongside the '320 in Affymetrix v. Illumina, No. 1:07‑cv‑00670‑JJF — e.g. US 6,420,169 and US 6,416,952 — appear in the same family chain (the WO 03/092881 family annex lists US 5,143,854 / 5,424,186 / 5,527,681 / 5,770,456 / 6,379,895 / 6,403,320 B1 / 6,416,952 / 6,420,169 / 6,506,558 / 6,610,482 together). Same-family members sharing the June 7, 1989 priority are not §102 prior art against the '320.

3. The references cited on the face of US 6,403,320 (as retrieved)

These are the entries the applicant/examiner listed in the "US Patent References" block of the '320 front page. FPO prints that block in descending patent-number order; my retrieval captured the top of the list (highest numbers = latest issue dates) and stopped at US 4,808,508, so the remainder (the lower-numbered, earlier references) is not in evidence here.

Foreign patent documents / other publications for the '320: not retrieved. (The JP 4‑262799 / JP 4‑304900 and WO 95/09248 entries that surfaced in my searches belong to a different patent's front page — US 9,637,784 — and I have deliberately not attributed them to the '320.)

4. §102 analysis, reference by reference

Critical caveat, stated once and applicable to every row: I could not obtain the '320's claims. I therefore express the anticipation analysis at the claim-feature level ("claims reciting X"), not with literal claim numbers. Assigning literal claim numbers without reading the claims is exactly the fabrication mode that prior sections of this analysis (correctly) refused to enter. The feature descriptions used below are drawn from (a) the '320 abstract in the authoritative text and (b) the summary of the invention as published in the sibling US 2004/0038268 A1 — which describes a composition comprising a plurality of positionally distinguishable sequence-specific reagents attached to a solid substrate … representing substantially all possible sequences of a preselected length, plus analyzing/sequencing methods. That sibling summary is inference about the '320's likely claim scope, not the '320's claim text.

Group A — Closest §102 candidates (the only ones I would call serious single-reference anticipation risks):

• US 4,833,092 (Geysen), issued 1989‑05‑23. This is the single most important date fact in the whole list: it issued 15 days before the June 7, 1989 critical date, so — assuming the June 7, 1989 priority is perfected — it is genuine §102(a) art, not merely §102(e) art. Geysen teaches synthesis of large numbers of peptides at positionally discrete, spatially addressed locations on solid supports, screened for specific binding. It potentially anticipates claims reciting a plurality of positionally distinguishable binding reagents attached to a solid support if such a claim is not limited to nucleic acids and does not require "substantially all possible sequences of said preselected length." Against claims reciting oligonucleotide probes or VLSIPS light-directed synthesis, Geysen is at best §103 combination art.
• US 4,877,745 (Hayes), issued 1989‑10‑31. Reagent fluid dispensing/printing with spatial addressing. §102(e) candidate (verify filing date) against any claim reciting depositing/delivering different reagents to different predefined regions of a substrate. Post-dates the critical date on its face, so it cannot be §102(a)/(b).
• US 4,923,901 (Koester), issued 1990‑05‑08. Membranes with bound oligonucleotides and peptides. §102(e) candidate (verify filing date) against claims reciting an oligonucleotide or peptide attached to a solid support. It does not, on its title/§112 disclosure as characterized, show an array of many different sequences at known positions.

Group B — Hybridization/capture methods (potentially relevant to method claims):

• US 4,925,785 (Wang) and US 4,923,901 (Koester) and US 4,921,805 (Gebeyehu) — all issue after 1989‑06‑07, so §102(e) only, contingent on filing dates. These are the natural §102 candidates against claims reciting hybridizing a target nucleic acid to support-bound probes and detecting the resulting duplex.
• US 4,888,278 (Singer) and US 4,886,741 (Schwartz) — issue late 1989; §102(e)-only. Directed to in situ hybridization in cells, i.e. a different sample format; more useful as background/§103 art than as §102 art.
• US 4,822,566 (Newman) and US 4,874,500 (Madou) — sensor arrays, but chemical/electrochemical sensors, not sequence-specific reagents. Not anticipatory of any array-of-probes claim; they are the type of reference an examiner cites to show "arrays were known."

Group C — Detection/labeling and chemistry supporting claims:

• US 4,868,103 (Stavrianopoulos), US 4,820,630 (Taub), US 4,931,384 (Layton), US 4,855,225 (Fung) — detection modalities (energy transfer, interactive labels, optical readout, labeled-oligo detection). Potentially relevant to claims reciting labeling the target and optically detecting binding.
• US 4,849,513 (Smith) — amino-functionalized deoxyribonucleoside phosphoramidites: the enabling chemistry for attaching synthetic oligonucleotides to a derivatized surface. Relevant to claims reciting attachment of probes to the substrate via a linker/amino group, but it is a chemistry reference, not an array reference.
• US 4,865,990 (Stead) and US 4,812,512 (Buendia) — functionalized supports. Same function: they support "derivatized substrates were known," not the array.

Group D — Instrumentation sub-combinations (scan optics, autofocus, alignment):

• US 4,846,552 (Veldkamp), US 4,844,617 (Kelderman), US 4,810,869 (Yabe), US 4,811,062 (Tabata), and US 4,811,218 (Hunkapiller).
• These are the classic Affymetrix-family front-page components: they appear on the same reference lists as US 5,143,854 / 5,424,186. Treat them as §103 combination art / background for any apparatus-type claim (scanner, confocal reader, alignment, autofocus, sequencing instrumentation). If the '320's claims are pure method/composition claims (as the abstract and the sibling summary suggest), these references are non-anticipatory and largely non-analogous.

Self/co-invented and same-family citations: none of the front-page references retrieved is a same-family member; the US 5,143,854 / 5,424,186 / 5,800,992 / 6,416,952 / 6,420,169 siblings appear in the family record, not in the '320's cited-reference block I retrieved. If any same-family member is listed on the un-retrieved portion of the front page, it cannot be §102 art against the '320 (shared June 7, 1989 priority and overlapping inventive entity).

5. What I could not verify (do not treat as negative findings)

1. The '320's verbatim claims — still unverified. This is the controlling gap; every "potentially anticipates" statement above is feature-level and conditional.
2. The remainder of the cited-reference list. Retrieval stopped at US 4,808,508. The earlier/higher-relevance references (typically the 4,3xx,xxx–4,6xx,xxx range in this family) were not retrieved.
3. Filing dates of the §102(e) candidates. I did not retrieve front-page filing dates, so I cannot confirm which of the post‑1989‑06‑07 references actually qualify under §102(e). This must be checked.
4. "Other publications"/non-patent literature and foreign patent documents on the '320 front page — not retrieved at all. For an Affymetrix VLSIPS family member, the NPL list is often the more probative prior art (it is where the SBH/sequencing-by-hybridization literature sits).
5. Whether the '320 was ever subject to a §102 rejection at all. It issued on 2002‑06‑11, consistent with allowance over the cited art — but I have not read the file history (09/684,377) and will not infer an allowance rationale.
6. Consistent with the earlier sections: any of these references bearing on the '320's claims is relevant only to pre‑expiration conduct, the patent being listed "Expired – Fee Related" with anticipated expiration 2009‑06‑07.

Bottom line

The references cited on the face of US 6,403,320 B1 that I could actually retrieve are a component-art list typical of the Affymetrix VLSIPS family: support/attachment chemistry (US 4,849,513; 4,812,512; 4,865,990; 4,923,901), hybridization and capture formats (US 4,925,785; 4,921,805; 4,888,278; 4,886,741), detection/labels (US 4,868,103; 4,820,630; 4,931,384; 4,855,225), sensor arrays (US 4,822,566; 4,874,500), spatial reagent dispensing (US 4,877,745), and scanner/optics sub-combinations (US 4,846,552; 4,844,617; 4,810,869; 4,811,062; 4,811,218).

The one reference that could function as a true §102(a) reference on the dates alone is US 4,833,092 (Geysen), issued 1989‑05‑23 — 15 days before the June 7, 1989 critical date — and it is the only one of the retrieved set directed to positionally addressed, solid-phase–synthesized binding reagents screened for specific binding. Everything else issued after June 7, 1989 and can therefore only be §102(e) art contingent on filing dates I have not verified, or §103 combination/background art.

I cannot responsibly name the specific '320 claims each reference anticipates, because I still do not have the '320 claim text. To finish this properly, pull (i) claim 1 and all independent claims of 6,403,320 from USPTO PatentCenter for application 09/684,377, and (ii) front-page filing dates for US 4,833,092; 4,877,745; 4,923,901; 4,925,785; 4,921,805; 4,888,278; and 4,886,741. With those two things this section can be converted from feature-level to literal claim-number mapping.

I'll try to retrieve the actual claim text and prior-art references before attempting any §103 analysis, since an obviousness opinion is meaningless without the claims.

I have partial priors. Let me pull the full cited-reference list and the claim text.

§103 Obviousness Analysis — U.S. Patent No. 6,403,320 B1

Preliminary contradictions and disclosure limits (flagged per operating rules)

1. The "Prior Art section of this page" does not contain prior art references. The Google Patents text supplied for US 6,403,320 renders only algorithmic prior-art keyword tags — "accordance; substrate; different; array; nucleic acids" — plus a keyword-frequency table. No "Patent Citations" / "Cited By" document list was rendered in the fetched text. I therefore cannot treat the supplied page as the prior-art record. To avoid fabricating references, I reconstructed the cited-art list from FreePatentsOnline's record for 6,403,320 (partially retrieved) and the co-family record for US 6,610,482 (same specification family). Caveat: a sibling's citation list is not automatically the '320 list; I mark below which items I verified in the '320 record.

2. Date inconsistency. The system context gives today as 2026‑09‑27; the task instruction gives April 26, 2026, and the previously generated section references June/July 2026 Federal Circuit materials. I flag this rather than silently pick one.

3. The claims are still not verbatim-verified. This is consistent with the prior generated section, which correctly refused to paraphrase claim scope. I will keep that discipline: the analysis below is provisional and keyed to claim categories recoverable from the shared specification's "Summary of the Invention" (US 2004/0038268 A1, identical title/specification family), not to claim language I have read. Any statement of scope should be re-run after pulling the granted claims from PatentCenter for application 09/684,377.

I. Legal framework and effective prior-art date

• Patent: US 6,403,320 B1 (granted 2002‑06‑11; filed 2000‑10‑05; priority 1989‑06‑07).
• Because it was filed before 2013‑03‑16 and claims pre‑AIA priority, pre‑AIA §103 applies (35 U.S.C. §103 as in force, with §103(c) common-ownership exemption for §102(e) art).
• §102(b) critical date: 1988‑06‑07; §102(a) window: 1988‑06‑07 → 1989‑06‑07. References published/patented in that window are §102(a) art.
• Critical point: US 5,143,854 (Pirrung et al.), US 5,424,186, US 5,800,992, US 6,610,482, US 6,573,424 are not prior art against this patent — they share the 1989‑06‑07 priority and (for most) common inventors/assignee. Any obviousness theory built on them is legally unavailable. A naive analysis will cite them; I expressly exclude them.
• The analysis below follows Graham v. John Deere (scope, differences, PHOSITA level, secondary considerations) and KSR Int'l v. Teleflex (rational-articulation catalogues: predictable combination of known elements, design incentives, market/technical demand).

II. The prior-art record actually available

A. Verified in the 6,403,320 record (FreePatentsOnline "US Patent References")

(The list is truncated after 4,820,630 in what I retrieved; additional earlier refs almost certainly exist.)

B. In the sibling specification record (6,610,482) — flagged as not confirmed for '320

US 4,833,092 (Geysen, "Method for determining mimotopes," 1989‑05‑23); US 4,731,325 (Palva, "Arrays of alternating nucleic acid fragments for hybridization arrays," 1988‑03‑15); US 5,002,867 (Macevicz, "Nucleic acid sequence determination by multiple mixed oligonucleotide probes"); US 5,202,231 (Drmanac, sequencing by hybridization); US 5,171,695 (Ekins); US 5,200,051 (Cozzette, electrode arrays); US 4,988,617 (Landegren, ligase detection); US 4,542,102 & 4,713,326 (Dattagupta, photochemical coupling of nucleic acids to solid supports); US 4,458,066 (Caruthers, solid-phase oligo synthesis); US 5,112,962 (Letsinger); US 4,639,211-style multiple-pin synthesis (Houghten, US 4,631,211, 1986‑12‑23); US 4,689,405 (Frank, simultaneous synthesis on solid phase); US 5,149,625 (Church).

C. Non-patent literature a POSITA would know (all pre‑1989‑06‑07)

• Geysen, Meloen & Barteling, PNAS 81:3998–4002 (1984) — spatially addressable arrays of hundreds of peptides synthesized by solid-phase chemistry, each at a known position, screened in parallel by specific binding (antibody); and Geysen et al., Mol. Immunol. 23:709 (1986).
• Houghten, PNAS 82:5131 (1985) — simultaneous multiple peptide synthesis ("tea-bag").
• Frank & Döring, Tetrahedron 44:6031 (1988) — parallel synthesis on cellulose paper discs at discrete positions.
• Southern, J. Mol. Biol. 98:503 (1975) — immobilized nucleic acid / hybridized labeled probe format.
• Bains & Smith, J. Theor. Biol. 135:303–307 (1988) — explicit proposal to determine sequence by hybridizing a target to a complete set of short oligonucleotides and decoding the hybridization pattern.
• Lysov et al., Dokl. Akad. Nauk SSSR 303:1508 (1988) — the same concept.
• Drmanac et al., Genomics 4:114–128 (Feb 1989) — theory and practice of sequencing by hybridization.
• Palva US 4,731,325 (1988‑03‑15) — an "array" of nucleic acid fragments for hybridization.

III. Level of ordinary skill (PHOSITA)

As of June 1989: a Ph.D. or M.S. in molecular biology/biochemistry/chemical engineering, or equivalent, with 2–3 years' experience, familiar with (a) solid-phase peptide and oligonucleotide synthesis (Merrifield; Caruthers phosphoramidite chemistry), (b) membrane/bead immobilization of biomolecules, (c) hybridization assay formats (Southern, dot/slot blot, in-situ), (d) immunoassay and radio/fluorescent detection, and (e) the then-emerging sequencing-by-hybridization literature.

IV. Claim categories analyzed (provisional, from the shared specification)

Recoverable from the discipline disclosure (sibling US 2004/0038268 A1, ¶¶[0011]–[0029], same family):

• Category A (composition): a plurality of positionally distinguishable sequence-specific reagents attached to a solid substrate, each capable of specifically binding a predetermined subunit sequence of a multi-subunit length of at least three subunits, the reagents representing substantially all possible sequences of that length (e.g., ~3,000 different sequences, or ≥25% of possible subsequences), at a density of ≥25 regions/cm² and often on a substrate <4 cm²; reagents being oligonucleotides (≥5 nt) or monoclonal antibodies.
• Category B (method of analysis): exposing a target polynucleotide/polypeptide to the Category-A composition; determining the pattern of positions that specifically interact; comparing that pattern to a reference pattern (fingerprinting / classification / mapping).
• Category C (sequencing a polymer): preparing reagents each specific for a subsequence of preselected length; positionally attaching each to one or more solid-phase substrates (including beads); combining with the target; determining which reagents interacted.
• Category D (sequencing a segment): combining a substrate bearing chemically synthesized, positionally distinguishable oligonucleotides with a target to form "high fidelity matched duplexes," then determining which reagents interacted.

Each category is addressed below with a specific combination, motivation, and legal rationale.

V. Combination 1 — Category A (arrayed "substantially all sequences" composition)

Primary: Geysen US 4,833,092 / Geysen PNAS 1984 + Geysen Mol. Immunol. 1986.
Geysen discloses exactly the architectural core: a plurality of positionally distinguishable binding reagents (peptides), each at a known address, synthesized in parallel by solid-phase chemistry, collectively representing a systematic (in mimotope work, near-complete) set of sequences of a chosen length, contacted with a target (antibody), after which the positions that specifically bound are read out. Geysen's multi-pin block is, in practical terms, one support bearing many positionally distinguishable regions — which is the "single substrate" embodiment of the disclosure. Geysen thus supplies elements (A) reagents, (B) positional distinguishability, (C) specific binding to a predetermined subunit sequence of ≥3 subunits, and (D) the "substantially all sequences" concept.

Secondary: Koester US 4,923,901 + Smith US 4,849,513 + Dattagupta US 4,713,326.
These supply the substitution of oligonucleotides for peptides and the chemistry to attach them: Koester expressly claims membranes having bound oligonucleotides and peptides; Smith provides amino-derivatized nucleosides for covalent support attachment; Dattagupta provides photochemical coupling of nucleic acids to solid supports.

Tertiary: Madou US 4,874,500 + Veldkamp US 4,846,552 + Layton US 4,931,384.
Supply the miniaturized, addressable detection/transducer array and miniaturized optics, supporting the density (≥25 regions/cm²) and small-footprint (<4 cm²) limitations.

Motivation / KSR rationales:

• Same problem, same solution architecture. Geysen's and the array-of-oligos' shared problem is "identify a target's linear information content by parallel interrogation with a systematically ordered library of short probes." Substituting nucleic acid probes for peptides was a predictable substitution of a known element for a known purpose (KSR), because Watson–Crick pairing is more deterministic than antibody–epitope binding — i.e., the substitution promised better, not merely equivalent, performance.
• Design incentive / market pressure: Sanger/Maxam–Gilbert sequencing throughput was the acknowledged bottleneck; Bains & Smith and Drmanac had already articulated the array-of-probes concept in print.
• Reasonable expectation of success: the enabling pieces (solid-phase oligo synthesis, membrane/bead attachment, fluorescence detection) were each independently reduced to practice before 1989.

VI. Combination 2 — Category C/D (sequencing by hybridization on a support)

Primary: Bains & Smith 1988 + Drmanac 1989 + Lysov 1988.
These three collectively disclose the method: expose a target nucleic acid to a set of short oligonucleotide probes, detect which probes hybridized, and decode the sequence from the hybridization pattern. Drmanac 1989 in particular supplies the practical reduction with immobilized probes and the algorithmic reconstruction.

Secondary: Koester US 4,923,901 / Dattagupta US 4,713,326 / Palva US 4,731,325 / Singer US 4,888,278.
Convert the solution-phase/two-dimensional-probe concept into a positionally addressable support-bound array ("which positions specifically interact" maps directly to the array readout). Palva's "arrays of alternating nucleic acid fragments for hybridization arrays" is a direct structural teaching of arrayed nucleic acids on a support.

Tertiary: Layton US 4,931,384 / Fung US 4,855,225 / Taub US 4,820,630.
Detection of hybridization at defined positions by optical/label means — supplying the "determining" step.

Motivation: Drmanac and Bains & Smith state the express objective (de novo sequencing without electrophoresis). Immobilizing the probe set on a support to gain parallelism, addressability, and reusability is the obvious design choice in light of Koester/Palva/Southern, and is driven by the same rationale that led Geysen to array peptides on a physical support. KSR "design incentive" and "predictable use of prior-art elements" rationales both apply. The bead embodiment (Category C) is squarely taught by classic solid-phase supports (Caruthers 4,458,066; polystyrene/agarose supports).

VII. Combination 3 — Category B (fingerprinting / classification / mapping by pattern comparison)

Primary: Singer US 4,888,278 (in-situ hybridization of many probes read by position in morphologically intact material — i.e., the sample's "source" classification) + Geysen (parallel specific-binding readout) + Southern 1975 (hybridize labeled target to immobilized sequences).

Secondary: Taub US 4,820,630 (interactive labels for nucleic-acid-sequence assays) and Stead US 4,865,990 (support chemistry) supply the labeling and support elements.

Motivation: The "compare to a reference pattern" step is the standard diagnostic/differential logic already used in in-situ hybridization and blot fingerprinting; applying it to a high-density array is an obvious application of a known technique to a known product, and the disclosure itself frames the benefit as automation/parallelism rather than any new biophysical principle.

VIII. Dependent/limitation-level obviousness

IX. §102 vs. §103 posture (important distinction)

For the sequencing-by-hybridization method categories, Bains & Smith (1988), Lysov (1988) and Drmanac (Feb 1989) are individually close to the claimed concept and create a §102 anticipation risk — if the claims do not positively require a physical, positionally addressable array on a support with optical/positional readout, those references (singly or in combination) may be more than merely "obviousness" art. Where the claims do require the array-on-support with positional determination, the combination of the SBH concept references with Koester/Palva/Dattagupta/Singer is the §103 theory. Any validity opinion must therefore separate the "algorithm" claims from the "apparatus/array" claims — which again requires the verbatim claim text.

X. Anticipated secondary considerations and rebuttals

• Unexpected results. The likely rebuttal is that mismatch discrimination at array scale was thought unattainable. Rebuttal: the patent's own specification concedes that "high fidelity matched duplex" formation was the design requirement — evidence the POSITA expected to engineer stringency, not evidence of surprise.
• Industry skepticism. Contemporaneous doubts about SBH accuracy exist in the literature; but skepticism must be directed at the claimed invention as a whole and must show the art taught away. The art (Bains & Smith, Drmanac) taught toward, not away.
• Licensing/commercial success (Affymetrix GeneChip, including the later assertion against Illumina's BeadArray in D. Del. 1:07‑cv‑00670). Commercial success and a nexus could be argued, but the scope of the nexus question is unanswered absent claim language; also note the patent reached anticipated expiration 2009‑06‑07 and is "Expired – Fee Related," so any obviousness challenge would have had to be resolved within the enforcement window (2002–2009).
• Common ownership. If any §102(e) reference was commonly owned with Affymetrix as of the invention, pre‑AIA §103(c) would disqualify it — must be checked reference-by-reference.

XI. Bottom line

On the record I can actually ground, the strongest §103 combinations are:

1. Geysen ('092 / PNAS 1984) + Koester ('901) + Smith ('513) + Bains & Smith (1988) or Drmanac (1989) [+ Madou '500 / Veldkamp '552 for density/miniaturization] → renders the arrayed-composition category obvious.
2. Bains & Smith (1988) + Drmanac (1989) + Palva ('325) or Koester ('901) + Layton ('384) / Taub ('630) → renders the sequence-determination-on-a-support method categories obvious.
3. Singer ('278) + Geysen + Southern (1975) + Taub ('630) → renders the fingerprinting/classification categories obvious.

The motivation in all three is the same and is strong: a POSITA facing the sequencing-throughput bottleneck had an express printed proposal to use complete probe sets (Bains & Smith; Drmanac), a demonstrated physical architecture for parallel positionally-addressed specific-binding arrays (Geysen), and mature chemistry for oligo synthesis and support attachment (Caruthers; Smith; Koester; Dattagupta). Under KSR, this is a textbook predictable combination of known elements with an articulated design incentive.

However — and this is the controlling caveat — I have not read the granted claims of 6,403,320. The analysis above is keyed to claim categories inferred from the shared family specification, not to claim language. Before this is used for anything operative:

1. Pull the claims from USPTO PatentCenter for application 09/684,377 (or the granted PDF), and confirm (a) the actual independent-claim count and (b) whether the independent claims are composition claims, method claims, or both.
2. Confirm the full "Patent Citations" list on the granted face (Google Patents' citation panel, or the printed patent), rather than the partial FreePatentsOnline extract and the 6,610,482 sibling list I had to use.
3. Verify each reference's §102 category and date against the 1989‑06‑07 priority, and run the pre‑AIA §103(c) common-ownership screen.
4. Separate SBH-algorithm claims (potential §102 exposure) from array-support claims (the §103 combinations above).

I have deliberately not supplied quotable claim language or a limitation-by-limitation chart, because doing so from memory would fabricate scope.