Question and Answer Sheet - USPTO Responses.pdf

Question # Reference Question Category

(Contract or Technical) Question

Answer

Patent vs. NPL Balance: The RFI mentions interest in US Patents, Foreign Patents, and Non-Patent Literature (NPL).

Technical For biosequences, Is there a priority on value-added, manually curated patent sequence data versus raw public repository data (like WIPO)?

Biosequences searching is based on the premise to search electronic sequence listing data submitted by applicants, and support searching of molecular sequences using data stored from both applicant submissions and public/commercial databases of published sequence information. Claimed biosequences are proprietary information. For the specific case of chemically-derived residues, there is a priority use for manually-curated data (i.e. STNext and CAS Registry) in order to search for information that is not typically searchable in other sequence databases and search tools, like NCBI BLAST and ABSS. Priority for sequence searching at USPTO is the in-house ABSS search system with EMBOSS implementation, which searches USPTO-generated databases as well as some commercial (e.g., Geneseq) and publicly available (GenEmbl) databases within an environment that is secured against public access.

Markush structure database including registry number identifiers of structure records for patents as well as Markush structures derived from INPI.

Markush structures of organic and organometallic molecules must be searchable and patent reference information must be displayable.

Technical Please can you clarify wht you mean by Markush structures derived from

INPI?

INPI (Institute National de la Propriete Industrielle) which has database coverage from 1961-1987.

Algorithm Specificity: The RFI mentions searching for 'sequences embedded in larger sequences' and 'exact sequences'.

Technical

Can you clarify the specific alignment algorithms required (e.g., BLAST, Smith-Waterman, MOTIF, Antibody CDR, Multiple Sequence Search) and if there are specific requirements for percentage identity/similarity scoring parameters?

Smith-Waterman, oligomer, score-over-length are a few algorithms used in our in-house database. Multiple algorithms and scoring tables are required: Smith-Waterman, ClustalW, Needleman-Wunsch, oligomer searches, p2n and n2p searches. BLAST (a public database) can be used by examiners, if they choose, through NCBI’s website, provided they are searching publicly-available data. BLAST is a faster, but not a rigorous, algorithm compared to Smith-Waterman. Does motif refer to the Registry motif searching? If so, that looks for exact matches, allowing for query variability (a feature that is available in Emboss searching as well). Oligomer searches look for the longest contiguous exact match with no gaps, and score-over-length is used to look for fragments within longer sequences.

Non-Standard Residues: Regarding the requirement to search 'claimed non-standard residues':

Technical Does the USPTO require a pre-defined dictionary of modified bases/amino acids, or user-definable custom residue definitions during the query setup? OR just based on ST.26

USPTO follows the international standard symbols for amino acids and nucleic acids as well as ambiguity symbols. Also, to address many claims, searchers need to be able to define variables at multiple positions (e.g., position 10 can be Ala, Leu, or Val and position 30 can be Asp or Glu)---currently, this can be done using EMBOSS Fuzzpro or motif searching in STN. Sequence listings are dictated by ST25 or ST26 rules based on the filing date.

Self-Hosting vs. SaaS: The RFI asks about 'self-hosting options'.

Technical Is there a security requirement that precludes a pure Multi-Tenant SaaS environment for the biosequence search platform?

No. FEDRAMP authorized would be preferred.

AI Implementation: The RFI specifically requests information on Artificial Intelligence and Generative AI. Technical

In the context of biosequences, is the USPTO interested in AI for query construction or for the analysis/summarization of search results?

Yes, we are interested in AI for query construction and analysis/summarization of search results.

None Given

Technical

Would you be willing to consider a two-product approach that together satisfies all of your requirements, rather than a single comprehensive solution? Clarifying further—does the RFI mandate an all-inclusive offering, or are you open to evaluating distinct products for different use cases?

Yes, multiple products that together provide a whole or part of the requirement will be considered.

RFI Attachment 2 - "Sequence Searching"

Technical Are DNA/RNA/protein sequence searches mandatory or optional?

Yes, having DNA/RNA/protein sequence searching are mandatory.

RFI Attachment 2 - "Bibliographic Searching"

Technical Which patent authorities and minimum coverage years are mandatory (USPTO, EPO, WIPO, CNIPA, JPO, etc.)? Should vendors provide full-text plus chemical structures for all jurisdictions?

Vendors should provide as many patent authorities as possible and years of coverage as possible. Vendors should provide as many text and structure for as many jurisdictions as possible.

RFI Attachment 1 - "Chemical reaction database…"

Technical For chemical reaction databases, what level of detail is required—conditions, yields, catalysts, atom mapping? Should the platform support reaction prediction or retrosynthesis planning?

Chemical reaction databases should include but not limit itself to the following: having the to ability to show/label reactants; to show the show/label the products; insert the yield symbol; identify bonds broken; cleavage bonds, reaction predictions, retrosynthesis planning, etc. Ability to look for reactions using Registry Numbers, functional groups, function group class terms, special reaction terms such as yields, etc.

RFI Attachment 2 - "Bibliographic Searching"

Technical Which controlled vocabularies (MeSH, ChEBI, CPC/IPC) must be supported for bibliographic searching? Should USPTO be able to customize or extend vocabularies?

Controlled vocabulary would be essential to text searching such as MeSH, ChEBI, CPC/IPC and other controlled vocabulary. Customizing or extending vocabulary terms is not necessary.

RFI Attachment 1 - "Directory, substance, and inventory database of regulated chemicals listings" Technical

Which regulatory inventories (TSCA, REACH, GHS, CWC, CFATS) are mandatory? Should vendors provide machine-readable crosswalks linking substances to these lists?

As many regulatory inventories would be ideal to have as possible for searching.

RFI Attachment 2 - "Bibliographic Searching" - "Users will be able to download a transcript limiting only to the publications of interest…" Technical

What file formats are acceptable for exporting search transcripts with associated structures (PDF, DOCX, XML)? Must exports include embedded identifiers (InChI, SMILES)?

As many file formats as possible such as Word Docs, PDFs are acceptable and others are acceptable. No it is not mandatory that exports include InChi or SMILES, but it would not hurt to have them.

RFI Attachment 2 - "Responses shall include…Licensing Model...Specify what each license type includes, how usage is measured and any limitations" Contract

How does USPTO define usage measurement—named seats, concurrent users, API calls, or data volume? Are automated usage reports required for auditing?

Minimal expectation is usage data (search, views, downloads) per user with timestamps. Preference is COUNTER compliant reports delivered through the SUSHI protocol.

RFI Part II

Contract Will we receive a response—positive or negative—to the Phase 1 proposal that includes feedback?

All respondents will receive a notification of the date and time for their Phase 2 product demonstration.

None Given

Contract

Could you clarify who the decision-makers are in this process and who will be evaluating the proposals and the demonstrations? For example, is the audience composed of scientists or end users, a review board, librarians, or another body?

Evaluation team will consist of end users (examiners and STIC searchers), librarians, and direct supervisors.

17 None Given Contract Is there a confirmed budget allocated for this purchase? The USPTO cannot discuss upcoming budget allocations.

None Given

Contract Who are the primary users of your chemical database subscription, and how many users do you have?

The primary users of the chemical database subscription are examiners in Technical Center 1600 (Biotechnology and Organic Chemistry), Technical Center 1700 (Chemical and Materials Engineering), and the Scientific and Technical Information Center (STIC) searchers. Total estimated users is between 2-3,000.