A biological sequence search compares a nucleotide or amino acid sequence with the sequences disclosed in patents, published applications and scientific literature. Alignment software scores every match, and the searcher reads the documents behind the strongest matches in order to decide whether the sequence is disclosed, claimed or both.
Keyword searching alone rarely finds this prior art, because a claim may identify a protein only by a SEQ ID number. Specialist biological sequence search work therefore starts from the residues themselves.
When a biological sequence search is used
The same search answers different questions at different stages of research and commercial work:
- patentability, before filing on a new antibody, peptide or enzyme;
- freedom to operate (FTO), before a sequence-based product is made or sold;
- invalidity, when a granted claim that recites a sequence is challenged;
- due diligence, before a licence or an acquisition.
Each purpose changes the scope of the search, mainly in which documents are read. An FTO search focuses on the claims of patents in force and of pending applications, as the freedom to operate search guide explains. A patentability search reads any earlier disclosure, following the method in the prior art search guide.
Where patent sequences are found
Most patent sequences sit in the sequence listings filed with applications. In the US, the current rule covers applications filed on or after 1 July 2022. An application that discloses a sequence of 10 or more specifically defined nucleotides, or 4 or more specifically defined amino acids, must include it in an XML sequence listing. The format follows Standard ST.26 of the World Intellectual Property Organization (37 CFR 1.831, checked October 2026). A separate post explains why sequence listings are needed in US applications.
| Source | What it holds | Typical use |
|---|---|---|
| Free patent sequence tools, such as Lens PatSeq | Patent sequences from several offices, searchable with BLAST | Early screening |
| Commercial platforms, such as STN, Orbit and GenomeQuest | Large patent sequence collections, some with literature sequences, plus search filters | Full patentability, FTO and invalidity work |
| Public sequence databases, such as GenBank and UniProt | Mainly research sequences; GenBank also holds patent sequences | Non-patent prior art and cross-checking |
Older documents need extra care, because some show the sequence only in a table, a figure or the description, and scanned text can change residues. Searchers therefore add a text search on SEQ ID references, protein names and gene symbols.
How a biological sequence search is scored
Most searches use the Basic Local Alignment Search Tool (BLAST) or a Smith-Waterman alignment. BLAST gains speed by skipping optimal scores for most database sequences, while Smith-Waterman computes an optimal local alignment for each one (BLAST tutorial). Each hit then carries several scores, which answer different questions:
- Percent identity is the share of aligned positions where the residues are the same.
- Query coverage is the share of the query that falls inside the alignment.
- E-value is the number of hits scoring at least as high that chance alone would produce in a database of that size. According to the BLAST FAQ, lower is more significant.
- Bit score is a normalised score with standard units.
Percent identity alone can mislead. A short fragment can match at full identity yet cover only a small part of a long query, so identity is read together with coverage.
Reading sequence claims against the results
Sequence claims are often written as ranges, such as a polypeptide with at least a stated percentage of identity to a numbered SEQ ID. The patent may define how identity is calculated, including how gaps are counted, and that method can give a different figure from the search tool. A biological sequence search therefore records every parameter and flags hits near a claimed threshold for the attorney.
Antibody claims raise a further point, since many define the antibody by its complementarity-determining regions (CDRs). These short regions are searched as separate queries, because a full-domain search may rank an exact CDR match low.
What a biological sequence search report contains
A biological sequence search report from a specialist firm usually contains the following items:
- the query sequences and parameters, such as algorithm and thresholds;
- the databases searched, with the date of each search;
- a results table with identity, coverage and E-value;
- the alignment of each key hit;
- bibliographic data, family members, legal status and a note on whether the sequence appears in the claims or only in the description.
Commissioning a biological sequence search
Before a biological sequence search starts, the search firm usually needs the sequences in FASTA or ST.26 format and the purpose of the search. It also needs the countries of interest and any claim language that defines identity. With these details it can agree databases and thresholds in advance.
For a quote, see our biological sequence search service or contact us.
This guide is general information and is not legal advice.
Frequently Asked Questions
How is a sequence search different from a keyword patent search?
A biological sequence search uses the residues as the query, while a keyword search uses words. Alignment scoring finds matches described under different names or only by SEQ ID number. Keyword searching still helps with older documents that have no listing.
What percent identity threshold is used in a sequence search?
There is no single threshold for every search, so the searcher sets it from the purpose and the claim language. It usually sits below the lowest identity that a relevant claim recites. Hits near the threshold are reviewed by hand, with coverage read alongside identity.
Can a sequence search find sequences in scientific literature?
Yes, in part. Public databases such as GenBank and UniProt hold sequences from research submissions and published studies, and these align in the same way as patent sequences. Some articles print sequences without a database record, so a literature text search is often added.
Does a sequence search cover the claims or the whole patent?
It covers the whole document, because listings include sequences from the description as well as the claims. The report states where each match appears, which matters most in an FTO search, where only claims in force, or pending claims that may be granted, can block a product.
What format should a query sequence be in?
FASTA is the usual format for a query sequence, because alignment tools accept it directly. Each record has a header line starting with a greater-than sign, followed by the residues. Sequences in an ST.26 listing can be converted to FASTA, keeping the SEQ ID numbers as labels.