SDC2 Biologics Sequence Review Report 2026: Similarity and Patent-Risk Signals
This SDC2 Biologics Sequence Review Report was produced with PatSnap Biology Modality MCP workflows. It connects patent-scale similarity search, sequence retrieval, pairwise alignment and antibody–antigen evidence in a reproducible screen. Explore the PatSnap MCP servers used in this report.
Review date: 17 September 2026. This report supports R&D and IP triage; it is not a legal opinion, freedom-to-operate conclusion or validity analysis. A sequence hit does not establish infringement. “Claimed” is a database annotation that still requires family consolidation, live-claim review and jurisdiction-specific claim construction.
Executive sequence-risk thesis
SDC2 is a protein directed to Syndecan-2. Public therapeutic sequence data from a reviewed UniProtKB/Swiss-Prot human protein record supplied the query material: Full-length protein 201 aa. PatSnap’s all-patent protein search returned a closest review hit at 100.00% query identity across 201/201; the returned record was marked No.
The screening conclusion is elevated diligence priority, not “blocked” or “clear.” Similarity is most informative when interpreted by domain location, construct architecture, target evidence and the language of live claims. Conserved receptor or scaffold regions can produce high percentages without proving asset-level collision, while a small number of substitutions in a binding epitope, engineered domain or junction can be disproportionately important.
Sequence scope and MCP workflow
The review began with a traceable reviewed human protein query. The full-length protein was submitted with ls_sequence_search_submit against ALLPATENT and CLAIMS protein records, allowing gaps, using E-value ≤1×10−3 and a 20-record task cap. Results were retrieved with ls_sequence_search_get_results; the closest informative hit was resolved with ls_sequence_fetch and compared using ls_sequence_alignment. ls_patent_sequence_fetch and ls_antibody_antigen_search supplied patent and target context.
Similarity search readout
| Query | Search volume | Leading evidence |
|---|---|---|
| Full-length protein (201 aa) | 20 returned records | 100.00% identity; 201/201; claimed: No |
The result set contained 20 records. These are search records, not unique inventions, enforceable claims or independent families. Publications can repeat the same sequence across jurisdictions, examples, comparators and continuations. Deduplication by earliest priority, applicant, simple family and legal status is required before business conclusions.
Module-level reading of the construct
The available query set contains one independently searchable full-length module. For a receptor or protein format, extracellular domains, transmembrane segments, intracellular regions, signal peptides and engineered junctions should be interpreted separately. A highly conserved domain may be shared across natural variants, constructs and assay reagents, while a shorter engineered segment can be more discriminating.
The search total also provides a practical specificity check. A query that reaches the task cap is less discriminating at the raw-sequence level than one with a small, concentrated result set. Reviewers should prioritize distinctive domains and junctions, then use widely conserved regions as corroboration. This ordering reduces the risk of treating platform-level or natural-sequence similarity as evidence of an asset-level collision.
Pairwise alignment: what the closest hit means
The selected full-length comparison reported 201/201 identical positions, query coverage 201/201, E-value 0 and 0/201 gaps. The matched subject sequence was resolved as sequence 1197325622, 214 aa.
Percentage identity alone is not a claim chart. The next review should map every mismatch to functional domains, binding interfaces, engineered residues, cleavage sites, transmembrane segments and any claimed SEQ ID. Fragment and domain-only queries should also be searched because a full-length comparison can obscure shorter protected modules.
Antibody–antigen and patent-sequence evidence
The exact-name antibody–antigen query returned 104 records. The leading patent record was WO2023215787A1, “Compositions and methods for selective regulation of vascular permeability”. Follow-up patent-sequence retrieval resolved 232 sequences. These records strengthen target-context evidence but do not establish that every listed sequence is part of a live claim.
Evidence strength and blind spots
The evidence is strongest for the exact query sequence, returned alignment coordinates and patent records directly resolved by the MCP tools. It is weaker where aliases are absent, where a publication supplies fragments instead of the administered construct, or where the result set is capped before all lower-ranked families are reviewed. Names can also refer to a parental molecule, assay reagent, comparator or natural antigen; the underlying sequence and target annotation must be read before attribution.
Patent sequence listings are deliberately broad and can contain screening libraries, consensus sequences, controls, antigens, primers and manufacturing elements alongside candidate therapeutics. Conversely, relevant claims may define a molecule through domain combinations, percentage-identity thresholds, functional binding language or genus formulas without reproducing the exact commercial sequence. This is why the report combines sequence similarity with antibody–antigen evidence and still stops short of a legal conclusion.
Patent-risk interpretation
| Dimension | Screening signal | Required next check |
|---|---|---|
| Sequence proximity | 100.00% closest reviewed identity | Consolidate families and map mismatches by domain. |
| Claim annotation | No | Read live independent and dependent claims in relevant jurisdictions. |
| Target evidence | 104 exact-name relations | Search aliases, development codes and each target separately. |
| Legal conclusion | Not determined | Assess priority, ownership, licensing, prosecution, validity and territorial status with counsel. |
Diligence actions
- Confirm that the reviewed sequence matches the clinical or commercial construct rather than an early lead or analytical reference.
- Repeat searches for domain-only queries, engineered junctions and nucleotide sequences, then cluster records by patent family.
- Annotate each high-identity hit by functional domain, binding interface, scaffold, transmembrane region and engineered residue.
- Retrieve the underlying sequence listing and verify the returned sequence number and “claimed” annotation against the filed document.
- Build jurisdiction-specific claim charts for live families and document licenses or platform rights affecting Syndecan-2 programs.
- Rerun the MCP workflow at the transaction or development decision date because sequence and legal-status coverage changes.
Bottom line
SDC2 has a traceable protein sequence. The combination of 100.00% closest-hit identity, No claim annotation, 104 exact-name antibody–antigen records and 232 directly fetched patent sequences supports elevated diligence priority. The defensible outcome is a prioritized, domain-aware family and claim review—not a binary clearance statement.
Sources and methodology notes
Query sequence: UniProtKB/Swiss-Prot reviewed human protein record. Core similarity, sequence, alignment, patent-sequence and antibody–antigen evidence was retrieved through PatSnap Biology Modality MCP on 17 September 2026. Database annotations and legal status can change; rerun at the decision date.
