RCSB PDB vs Taiwan Health MCP
A side-by-side look at the RCSB PDB and Taiwan Health MCP MCP servers: transport, auth, tools, quality signals, and the config to install either one in Claude, Cursor, Windsurf or VS Code.
Verdict: choose RCSB PDB for biology & bioinformatics work as a local process; choose Taiwan Health MCP for biology & bioinformatics. You can also run both side by side in one mcpServers block.
Which MCP server should you choose?
Choose RCSB PDB when
- You need dedicated capabilities in the Biology & Bioinformatics domain.
- You prefer a local stdio subprocess transport.
- You want tools like
rcsb_render_report, rcsb_find_go_terms, rcsb_find_interpro_domains.
Choose Taiwan Health MCP when
- You need dedicated capabilities in the Biology & Bioinformatics domain.
- Its auth model fits you: No auth required.
- You have the keys it needs:
POSTGRES_PASSWORD. - You want tools like
health_check, calculate_chads2_vasc, calculate_wells_pe.
Feature & specification comparison
Rows where the two servers differ are marked.
| Spec | ||
|---|---|---|
| Category | Biology & Bioinformatics | Biology & Bioinformatics |
| Quality signal | 60/100 · Good | 60/100 · Good |
| Install command | Confirmed | Not confirmed yet |
| Transport | Local subprocess (stdio) | — |
| Auth | Not declared | No auth required |
| Pricing | Not declared | Free |
| Required env vars | None listed | POSTGRES_PASSWORD |
| Verified owner | Community listing | Community listing |
| GitHub stars | 11 | 167 |
| Directory activity | 1 views 0 installs 0 upvotes | 2 views 0 installs 0 upvotes |
| Tools | 36 | 59 |
| Full listing | View RCSB PDB | View Taiwan Health MCP |
Tools & capabilities
RCSB PDB tools 36
rcsb_render_reportRender a structure-search report as a styled, self-contained HTML page. Call this LAST, after the searches and rcsb_get_* calls that produce the values shown in the report. Supply facts only: the page is rendered from a fixed server-side template, so do not write HTML yourself and do not rewrite what this tool returns. Key fields of ``report``: title: page heading describing the search. api_calls: one per Search/Data/Sequence-Coordinates call, using the ``editor`` object the tool returned VERBATIM. Resolver and discovery tools have no editor link — pass label and tool_name only. result_type: what your ids ARE — "entry" for PDB entry ids (4HHB, the default) or "ligand" for chemical component ids (ATP). It must match the ids you send: a mismatch resolves nothing, and every derived value comes back empty. results: the identifiers you are reporting, in the order you want them ranked, each ``{"id": ..., "evidence": {...}}``. Evidence has two fields: ``grounds`` (the tool-returned value the match rests on) and optional ``interpretation`` (your own reading of it). Splitting them is what stops your inference from being read as something the archive returned. data_usage: ordered narrative of how each call shaped the final set; each item's ``body`` is plain prose — your own account of how you worked. Returns: RenderReportResult with EITHER a `url` or `html` (never both), plus template_version. Prefer `url` — it is a self-contained link that renders the report on demand; deliver it to the user as-is. `html` is only returned as a fallback; write it to a `.html` file.rcsb_find_go_termsResolve a free-text molecular function, biological process, or cellular component / location (e.g. kinase activity, ATP binding, DNA repair, apoptosis, signal transduction, mitochondrial membrane, nucleus) to Gene Ontology (GO) terms, so you can run precise GO-based PDB searches instead of keyword guessing. Use this when a request involves what a protein DOES or where it acts — "proteins that <do X> / are involved in / participate in / are responsible for ...", "localized to / located in ...". Resolve the phrase to a GO id here, then filter on it with rcsb_query_attribute: exact_match on rcsb_polymer_entity_annotation.annotation_lineage.id, id AS A STRING ("GO:0004672"). The *_lineage.id paths are HIERARCHICAL — they match the term AND everything beneath it; use `in` with several ids to broaden. For ONLY that exact term without descendants use annotation_id instead. Results are based on how each entry's text matched your query. Broader, synonym, or differently-worded terms for the same concept may produce different results. Rank is unrelated to how much of the archive each term covers. Returns: {query, namespace, count, terms:[{id, name, aspect, pdb_entry_count?}]}.rcsb_find_interpro_domainsResolve a free-text protein domain, family, or fold (e.g. SH2 domain, immunoglobulin fold, zinc finger, beta-barrel, WD40 repeat, kinase domain) to InterPro entries, for precise InterPro-based PDB searches instead of keyword guessing. Use this whenever a request references a protein DOMAIN, FAMILY, or fold — "structures containing / with a <domain>", "<domain>-containing proteins", "members of the <family> family". Resolve the phrase to an accession here, then filter on it with rcsb_query_attribute: exact_match on rcsb_polymer_entity_annotation.annotation_id, accession AS A STRING ("IPR000719"). Lineage paths are not available; broaden with `in` over several accessions instead (IPR and PF mix freely). Ids come from InterPro ("IPR000719") or Pfam ("PF07859") — `source_database` on each entry says which. Both filter on the same attribute; nothing else needs to change. Results are based on how each entry's text matched your query. Broader, synonym, or differently-worded terms for the same concept may produce different results. Rank is unrelated to how much of the archive each term covers. Returns: {query, entry_type, count, entries:[{id, name, type, source_database, matched_text?, pdb_entry_count?}]}. `matched_text` appears only when the entry's NAME does not contain your words — it quotes the text that did match.rcsb_find_enzyme_classesResolve a free-text enzyme, enzyme class, or catalyzed reaction (e.g. alcohol dehydrogenase, protease, kinase, DNA polymerase, hydrolase, oxidoreductase) to Enzyme Commission (EC) numbers, for precise EC-based PDB searches instead of keyword guessing. Use this when a request references an enzyme, enzyme class, or reaction — including "enzymes that catalyze / break down / degrade / synthesize / hydrolyze / phosphorylate ...". Resolve the phrase to an EC number here, then filter on it with rcsb_query_attribute: exact_match on rcsb_polymer_entity.rcsb_ec_lineage.id, EC number AS A STRING. The *_lineage.id paths are HIERARCHICAL — they match the term AND everything beneath it; use `in` with several ids to broaden. A partial EC like "3.4.21" therefore matches the whole sub-subclass. Results are based on how each entry's text matched your query. Broader, synonym, or differently-worded terms for the same concept may produce different results. Rank is unrelated to how much of the archive each term covers. Returns: {query, count, enzymes:[{ec, name, pdb_entry_count?}]}.rcsb_find_disease_termsResolve a free-text disease, disorder, syndrome, or condition (e.g. diabetes, cancer, Alzheimer disease, cystic fibrosis) to MONDO ontology ids, for precise disease-based PDB searches instead of keyword guessing. Use for ANY request mentioning a disease/disorder/syndrome/condition — "structures involved in / associated with / linked to <disease>", "proteins implicated in <disease>". Resolve the phrase to a MONDO id here, then filter on it with rcsb_query_attribute: exact_match on rcsb_uniprot_annotation.annotation_lineage.id, id AS A STRING ("MONDO:0005148"). The *_lineage.id paths are HIERARCHICAL — they match the term AND everything beneath it; use `in` with several ids to broaden. A MONDO id therefore finds the disease and its subtypes. Results are based on how each entry's text matched your query. Broader, synonym, or differently-worded terms for the same concept may produce different results. Rank is unrelated to how much of the archive each term covers. Returns: {query, count, diseases:[{id, name, matched_synonym?, pdb_entry_count?}]}. `matched_synonym` appears only when the formal MONDO name does not contain your words — it names the synonym that did match.rcsb_find_organismsResolve a free-text organism, common name, or clade (e.g. human, mouse, baker's yeast, Escherichia coli, mammals, bacteria, primates) to NCBI Taxonomy ids, for precise taxonomy-based PDB searches instead of keyword guessing. Use when a request restricts structures by SOURCE ORGANISM or any higher taxon — a common name you want as a canonical taxon ("human", "fruit fly"), or a CLADE, which a plain name search cannot expand. Resolve the phrase to a taxon id here, then filter on it with rcsb_query_attribute: exact_match on rcsb_entity_source_organism.taxonomy_lineage.id, id AS A STRING ("9606", not 9606 — a bare number does not match). The *_lineage.id paths are HIERARCHICAL — they match the term AND everything beneath it; use `in` with several ids to broaden. A clade id ("40674" = Mammalia) therefore finds every organism beneath it; for a known exact species ncbi_scientific_name exact_match also works. An informal, polyphyletic group ("filamentous fungi", "extremophiles", "algae") is NOT a taxon and has no id: resolve the nearest CONTAINING taxon, then classify each hit from the lineage its own record returns. Results are based on how each entry's text matched your query. Broader, synonym, or differently-worded terms for the same concept may produce different results. Rank is unrelated to how much of the archive each term covers. Returns: {query, count, taxa:[{tax_id, scientific_name, common_name, rank, pdb_entry_count?}]}.rcsb_query_fulltextBuild a FREE-TEXT keyword query (e.g. "CRISPR Cas9", "hemoglobin"). Best for broad or exploratory lookups. When a request resolves to a clear attribute and value, prefer rcsb_query_attribute — more precise, and it avoids spurious keyword matches. BEFORE keyword-searching a biological CONCEPT (disease/function/domain/ enzyme/organism), resolve it to an ontology id with the matching rcsb_find_* tool and filter on the annotation instead; fall back to keywords only if that yields nothing. Matching spans many text attributes beyond the title — abstracts, keywords, synonyms, author names — so judge each hit yourself: a high `score` is text-relevance, NOT biological importance; never tell the user one hit is better than another because its score is higher. For evidence on a hit, rcsb_get_entries returns its title and, for many entries, a PubMed abstract (pubmed.rcsb_pubmed_abstract_text). Broader, synonym, or differently-worded terms for the same concept may produce different results. Returns: A query document — pass it to rcsb_search_request to run it, or to rcsb_query_composer to combine it with other queries first.rcsb_list_pdb_search_attributesDiscover the RCSB PDB Search schema: attribute paths, value types, and operators. Call this FIRST when a request resolves to a clear attribute and value but you don't know the exact path; pick the attribute here, build the condition with `rcsb_query_attribute`, then execute with `rcsb_search_request`. Returns: {count, match_mode, attributes, note?}. `attributes` holds {attribute, type, operators, description, enum?, nested_group?} records — the RCSB/PDB attribute path (e.g. "rcsb_entry_info.resolution_combined"), its value type (string/number/integer/date), the operators it supports (exact_match, greater, range, exists, ...), and a human-readable description. `enum` appears on the ~16% of attributes that accept only a FIXED SET of values (e.g. exptl.method); when it does, use one of those values verbatim — anything else is rejected by rcsb_query_attribute. `nested_group` appears on the ~19% stored in NESTED DOCUMENTS — an entry has many citations, an entity many binding affinities — and its value is the container path. For these, grouping selects the semantics: conditions sharing a nested_group built in ONE rcsb_query_attribute call with nothing else in it must hold on the SAME record; in separate calls each is matched independently. Group the ones that describe one record, leave the rest apart. The value is NOT the attribute's first path segment for all of them, so read it rather than deriving it. `match_mode` is "exact" (every word matched, best match first), "partial" (only some did — read `note`), "none" (nothing matched — read `note`), or "all" (query omitted, whole catalog).rcsb_query_attributeBuild a STRUCTURED query from attribute conditions — the precise alternative to keyword search, and preferred whenever a request resolves to clear attribute(s) and value(s). NEVER invent, guess, or infer an attribute path: call rcsb_list_pdb_search_attributes first if you don't know it or its operators. For a biological concept, resolve it to an ontology id with the matching rcsb_find_* tool and filter on that annotation. If a resolver returns no usable id, or a concept filter yields no hits, fall back to rcsb_query_fulltext for the concept. For ordinary constraints — resolution, organism, dates — an empty result is a valid answer: report it, don't keyword-search instead. An attribute path also works as the `query` for rcsb_describe_data_object: every attribute here is also a Data API field, so pasting one in tells you which rcsb_get_* tool fetches the value. Example ("human X-ray structures better than 2 A"): attributes=[ {"attribute": "rcsb_entity_source_organism.ncbi_scientific_name", "operator": "exact_match", "value": "Homo sapiens"}, {"attribute": "exptl.method", "operator": "exact_match", "value": "X-RAY DIFFRACTION"}, {"attribute": "rcsb_entry_info.resolution_combined", "operator": "less", "value": 2.0}, ] Returns: A query document — pass it to rcsb_search_request, or to rcsb_query_composer. Note the per-hit `score` of a pure attribute filter is near-uniform and carries NO biological meaning; don't rank hits by it.rcsb_query_sequenceBuild a SEQUENCE-SIMILARITY query (MMseqs2, BLAST-like) from a raw sequence. Use when you have actual residues to match. For a short motif or pattern use rcsb_query_seqmotif; for a named protein use rcsb_query_fulltext or a resolver. Returns: A query document — pass it to rcsb_search_request, or to rcsb_query_composer.rcsb_query_chemicalBuild a CHEMICAL query from a SMILES/InChI descriptor or a molecular formula. Use for ligand and small-molecule questions where you have the chemistry itself. To find a ligand by NAME, use rcsb_query_fulltext or filter on a chemical attribute. Returns: A query document — pass it to rcsb_search_request, or to rcsb_query_composer.rcsb_query_structureBuild a 3D SHAPE-SIMILARITY query against an existing PDB structure. Whole-shape similarity — use it for "structures shaped like X" / "same fold as X". For a geometric arrangement of specific residues use rcsb_query_strucmotif; for sequence similarity use rcsb_query_sequence. Returns: A query document — pass it to rcsb_search_request, or to rcsb_query_composer.- +24 more on the listing page
Taiwan Health MCP tools 59
health_checkReturn runtime readiness of the MCP server and every module-backed service. Call this first before any workflow to confirm the required services are online.calculate_chads2_vascCHA₂DS₂-VASc stroke-risk score for non-valvular atrial fibrillation (0–9). Accepts raw age/sex and computes brackets internally; also returns the 2024 ESC CHA₂DS₂-VA adjusted score (sex removed). Decision support only.calculate_wells_peWells criteria for pulmonary embolism (two-tier + three-tier risk). Binary findings are booleans; heart_rate_bpm is raw and the >100 criterion is computed.calculate_wells_dvtWells criteria for deep-vein thrombosis (two-tier + three-tier risk). All items are binary clinical findings.calculate_das28DAS28 rheumatoid arthritis disease activity. formula=esr needs esr (mm/h); formula=crp needs crp (mg/L). Joint counts are 28-joint (0–28); global_health is a 0–100 VAS (not the 0–10 scale CDAI uses). Returns score and activity band (remission/low/moderate/high).calculate_cdaiCDAI (lab-free) rheumatoid arthritis disease activity (0–76). Tender and swollen counts use the 28-joint assessment (0–28 each); patient and provider global assessments are 0–10.calculate_bmiBody mass index with Taiwan (Asian) cutoffs: <18.5 underweight, 18.5–23.9 normal, 24–26.9 overweight, ≥27 obese. Height in cm (not metres), weight in kg.calculate_egfrEstimated glomerular filtration rate using the CKD-EPI 2021 creatinine equation (race-free), adults ≥18 only. serum_creatinine_mgdl in mg/dL (convert µmol/L by dividing by 88.4). Returns eGFR in mL/min/1.73m² and KDIGO CKD stage.calculate_has_bledHAS-BLED bleeding risk on oral anticoagulation (0–9). Accepts raw age (the >65 criterion is computed) and raw systolic BP (the H criterion is computed as SBP >160 mmHg). Returns score and high/low bleeding-risk category (≥3 high).calculate_curb65CURB-65 community-acquired pneumonia severity (0–5). BUN in mg/dL (threshold >19), respiratory rate ≥30, hypotension SBP<90 or DBP≤60, confusion, age ≥65.calculate_qsofaqSOFA sepsis screening (0–3): respiratory rate ≥22, altered mentation (GCS <15), systolic BP ≤100. ≥2 suggests sepsis and further evaluation.calculate_gcsGlasgow Coma Scale (3–15). eye_response 1–4, verbal_response 1–5, motor_response 1–6. Returns total, E/V/M sub-scores, and mild/moderate/severe severity (≤8 severe).- +47 more on the listing page
Ready-to-paste client configs
Paste either block (or both) into your client config, e.g. claude_desktop_config.json or ~/.cursor/mcp.json. We only show a config when the install command is confirmed.
RCSB PDB
{
"mcpServers": {
"rcsb-pdb-2": {
"command": "uvx",
"args": [
"rcsb-mcp"
]
}
}
}
Taiwan Health MCP
No confirmed install command for Taiwan Health MCP yet — follow the setup steps linked from its listing page rather than a guessed package name.
Frequently asked questions
RCSB PDB is categorized under Biology & Bioinformatics and uses a local stdio subprocess. Taiwan Health MCP belongs to Biology & Bioinformatics and uses an install path we have not confirmed yet. Pick RCSB PDB for biology & bioinformatics work and Taiwan Health MCP for biology & bioinformatics.
Copy the mcpServers block from the config section on this page into your client's config (for example claude_desktop_config.json or ~/.cursor/mcp.json), then fully restart the client. Taiwan Health MCP has no confirmed install command yet — follow its own setup instructions.
RCSB PDB has no declared pricing — check the project's own site. Taiwan Health MCP is listed as free. Auth: no auth required. If a server needs environment variables such as API keys, define them under the env key in your client's MCP config.
Yes. MCP clients support multiple servers: add rcsb-pdb-2 and taiwan-health-mcp as separate keys in the same mcpServers object, and your assistant will call tools from either one.
On AllMCPs, RCSB PDB has a quality signal of 60/100 (Good) with 1 views, 0 installs and 11 GitHub stars. Taiwan Health MCP has 60/100 (Good) with 2 views, 0 installs and 167 GitHub stars.