RFI Attachment 2 - Biosequence Search SOO.pdf

Draft Statement of Objectives (SOO)

Next-Generation Super Intelligence (SI)-Assisted Biosequence Search System

1. Purpose

The United States Patent and Trademark Office (USPTO) seeks to acquire and implement a next-generation, commercially available biological sequence search system. The objective is to utilize Super Intelligence (SI)-assisted capabilities to modernize the architecture, increase the efficiency, and improve the quality of biological sequence searches (e.g., DNA, RNA, proteins, peptides, antibodies) conducted during patent examinations.

2. Background

The USPTO examines patent applications to determine whether the claimed invention satisfies U.S. patentability requirements. Biotechnology examination often involves applications that disclose biological sequences in a sequence listing and recite those sequences in the claims.

A claim may be narrow, such as one exact listed sequence, or broad, such as a sequence having at least a stated percent identity to one of many listed sequences. Claims may also cover fragments, fused sequences, variants, alternative residues, and related structural forms. As a result, sequence examination is not simply a document search.

Automated Biotechnology Search System (ABSS) is the USPTO’s current sequence-search system. It supports searching nucleotide and amino acid query sequences against a set of patent-sequence and non-patent-sequence databases. The examiner may run a basic search directly through the self-service ABSS interface or a more customized search executed by internal search support resources. ABSS currently employs the Smith- Waterman algorithm and performs sequence searches on information from various database including GenBank, GeneSeq, and Swiss-Prot, then stores the search results as exports to other USPTO systems integrated with ABSS.

The USPTO requires a modernized, cloud-based platform that reduces examiner burden, improves search focus, and leverages SI to assist in strategy generation and result prioritization while keeping the examiner as the ultimate decision-maker.

3. Scope

The scope of this acquisition includes the software, platform, integration, and SI capabilities necessary to provide end-to-end biosequence search and alignment. This includes ingesting patent documents (claims, specifications, sequence listings), executing

Draft complex searches across massive biological databases, and providing SI-driven analysis and summarization of the results in a secure, USPTO-hosted AWS cloud environment.

4. Core Objectives

The Contractor shall propose a solution and methodology that achieves the following high-level objectives:

i. Objective 1: Provide Comprehensive Core Biosequence Search Capabilities

a) Deliver highly accurate one-to-one nucleotide and protein alignment searching against large databases.

b) Support local and global alignments utilizing standard, publicly reviewable algorithms (e.g., Smith-Waterman, BLAST, Needleman-Wunsch, Hidden Markov Models).

c) Enable complex searches including fragments, patterns, regular expressions, wildcard positions, gaps, repeats, and reverse complement nucleotide searching.

d) Searches including:

a. nucleotide-to-nucleotide

b. nucleotide-to-protein

c. protein-to-protein

d. protein-to-nucleotide

e) Reverse complement nucleotide searching and clear distinction of results.

f) Ranking, filtering, sorting, and saving hits by multiple metrics, including score-based and query-match-based methods.

g) Provide robust sequence analysis tools, including pairwise/multiple sequence alignments, consensus sequence generation, and a comprehensive query sequence editor.

h) Retrieval, storage, and export of results in defined and customer-configurable formats.

ii. Objective 2: Optimize Search Strategy via SI Assistance

a) Utilize SI to analyze ingested case files (claims, specifications, sequence listings) and extract search-relevant sequence features.

b) Generate customized, optimal search strategies and query groupings based on the extracted features.

c) Provide structured, explainable recommendations to examiners to speed up the planning step and reduce inconsistency in search execution.

d) Search progress visibility with estimated completion times.

e) Access to completed result files for viewing and download.

f) Historical retrieval of prior search results.

g) Ability to fetch sequences and fragments from patent-related sequence sources.

h) Ability to filter, sort, search text, and export result sets.

i) Integration support for downstream USPTO workflows and related systems.

iii. Objective 3: Generate Quality Search Result Assessment and Prioritization

a) Utilize SI to organize, cluster, and prioritize search results in examiner-centered terms.

b) Group duplicative or related hits and link sequence hits directly to underlying patent or publication references.

c) Produce concise, explainable summaries detailing why specific references were elevated for review.

d) Support an iterative search approach by suggesting follow-up search strategies based on initial results.

iv. Objective 4: Incorporate Human-in-the-Loop Feedback

a) Design the system to allow active human feedback in SI/Machine Learning (ML) workflows.

b) Ensure the examiner can easily accept, reject, or revise SI-generated recommendations and search strategies.

c) Preserve a clear audit trail and record of user actions to enhance accuracy, safety, and ethical reasoning over time.

v. Objective 5: Modernize Architecture and Platform

a) Deliver a modular, maintainable, and cost-efficient bioinformatics platform that provides predictable, high-throughput search performance at scale.

b) Utilize cloud-native infrastructure (containerized, microservice-based) with automated scaling, fault isolation, and automated recovery.

c) Provide automated, configurable ingestion pipelines to import and refresh data from multiple internal and external sequence databases.

d) Integrate seamlessly with existing USPTO workflows and related systems through documented APIs.

5. Current Challenges:

a) Labor-intensive sequence-related examination for users– Examiners must manually convert complex patent disclosures into technical search instructions and then manually convert technical search output into examination-ready prior art analysis.

b) Sequence data volume continues to increase – As searchable data grows, poorly targeted searches can generate more noise and consume more examiner time.

c) Search quality and search efficiency are closely linked – A better-focused search is more likely to surface the most relevant art and less likely to burden the examiner with large volume of low-value results.

d) Operational and compute-resource considerations – Broad searches across very large nucleotide databases can increase cost and strain shared resources.

6. Operating Constraints and Security Requirements

The proposed solution must operate within the following constraints:

a) Hosting: The solution shall be cloud-based and self-hosted within the USPTO AWS cloud environment.

b) Security & Compliance: The solution must comply with FISMA, FedRAMP, FIPS, and NIST Special Publication 800-Series requirements to achieve and maintain an Authority to Operate (ATO). It must also comply with Government supply chain risk assessment (SCRA) requirements.

c) Security, Privacy, and Compliance: Maintaining FedRAMP authorization, integrating with USPTO Identity, Authentication, and Authorization Management (IAAM) via Okta, ensuring SI transparency (per EO 13960 and 14179), and enforcing strict data handling protocols (e.g., immediate session data deletion and automated export redaction).

d) Data Protection: The solution will interact with confidential patent data under 35 U.S.C. 122. Any internal SI tool enhancements must not be directly accessible from the internet.

e) Data Rights: The USPTO shall maintain sole rights to all proprietary data, models, technologies, and intellectual property developed or accessed related to this contract (excluding publicly available database data).

f) Accessibility: The solution shall be fully Section 508 compliant.

g) SI Principles: The tool must assist, not replace, examination judgment. All SI recommendations must be explainable, and the examiner must remain the final decision-maker.

SOO Attachments

Appendix A - USPTO Systems, Workflow Tools, and Searchable Data Sources

Appendix B - Current-State Workflow Tables

Appendix C - Future-State Workflow Tables

Appendix D - Pain Points and Capability Gaps