|
| 1 | +import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js"; |
| 2 | +import { z } from "zod"; |
| 3 | +import { contentClient } from "../clients/content.js"; |
| 4 | +import type { SearchResult, Event, Pathway } from "../types/index.js"; |
| 5 | + |
| 6 | +/** |
| 7 | + * Strip HTML tags from search result text. |
| 8 | + */ |
| 9 | +function stripHtml(text: string): string { |
| 10 | + return text.replace(/<[^>]*>/g, ""); |
| 11 | +} |
| 12 | + |
| 13 | +/** |
| 14 | + * Search Reactome, fetch details for the top result, and build |
| 15 | + * a human-readable explanation entirely from the Reactome data. |
| 16 | + * No external LLM API required. |
| 17 | + */ |
| 18 | +async function buildExplanation( |
| 19 | + query: string, |
| 20 | + detailLevel: "brief" | "standard" | "detailed" |
| 21 | +): Promise<string> { |
| 22 | + // Step 1: Search Reactome |
| 23 | + const searchResult = await contentClient.get<SearchResult>("/search/query", { |
| 24 | + query, |
| 25 | + rows: 10, |
| 26 | + cluster: true, |
| 27 | + }); |
| 28 | + |
| 29 | + const entries = searchResult.results.flatMap((group) => group.entries); |
| 30 | + |
| 31 | + if (entries.length === 0) { |
| 32 | + return [ |
| 33 | + `## No Results Found`, |
| 34 | + "", |
| 35 | + `Reactome has no entries matching **"${query}"**.`, |
| 36 | + "", |
| 37 | + "**Tips:**", |
| 38 | + "- Try a gene symbol like `TP53` instead of a full name", |
| 39 | + "- Try a pathway name like `apoptosis` or `cell cycle`", |
| 40 | + "- Check spelling with the `reactome_search_spellcheck` tool", |
| 41 | + ].join("\n"); |
| 42 | + } |
| 43 | + |
| 44 | + const lines: string[] = []; |
| 45 | + const topEntry = entries[0]; |
| 46 | + |
| 47 | + // Step 2: Fetch detailed info for the top result |
| 48 | + let detailed: Event | null = null; |
| 49 | + if (topEntry.stId) { |
| 50 | + try { |
| 51 | + detailed = await contentClient.get<Event>( |
| 52 | + `/data/query/enhanced/${encodeURIComponent(topEntry.stId)}` |
| 53 | + ); |
| 54 | + } catch { |
| 55 | + // If detailed fetch fails, continue with search data only |
| 56 | + } |
| 57 | + } |
| 58 | + |
| 59 | + // Step 3: Build the explanation header |
| 60 | + const name = detailed?.displayName || stripHtml(topEntry.name); |
| 61 | + lines.push(`## ${name}`); |
| 62 | + lines.push(""); |
| 63 | + |
| 64 | + // Basic metadata |
| 65 | + if (detailed) { |
| 66 | + lines.push(`| Property | Value |`); |
| 67 | + lines.push(`|----------|-------|`); |
| 68 | + lines.push(`| **Reactome ID** | ${detailed.stId} |`); |
| 69 | + lines.push(`| **Type** | ${detailed.schemaClass} |`); |
| 70 | + if (detailed.speciesName) { |
| 71 | + lines.push(`| **Species** | ${detailed.speciesName} |`); |
| 72 | + } |
| 73 | + if (detailed.isInDisease) { |
| 74 | + lines.push(`| **Disease pathway** | Yes |`); |
| 75 | + } |
| 76 | + if (detailed.hasDiagram) { |
| 77 | + lines.push(`| **Has diagram** | Yes |`); |
| 78 | + } |
| 79 | + lines.push(""); |
| 80 | + } |
| 81 | + |
| 82 | + // Step 4: Add the summary/description |
| 83 | + if (detailed?.summation && detailed.summation.length > 0) { |
| 84 | + const fullSummary = detailed.summation[0].text; |
| 85 | + // Strip HTML from summation text |
| 86 | + const cleanSummary = stripHtml(fullSummary); |
| 87 | + |
| 88 | + if (detailLevel === "brief") { |
| 89 | + // Just first 2 sentences |
| 90 | + const sentences = cleanSummary.match(/[^.!?]+[.!?]+/g) || [cleanSummary]; |
| 91 | + lines.push("### Summary"); |
| 92 | + lines.push(sentences.slice(0, 2).join(" ").trim()); |
| 93 | + } else { |
| 94 | + lines.push("### Description"); |
| 95 | + lines.push(cleanSummary); |
| 96 | + } |
| 97 | + lines.push(""); |
| 98 | + } |
| 99 | + |
| 100 | + // Step 5: For standard/detailed — fetch sub-pathways and participants |
| 101 | + if (detailLevel !== "brief" && detailed?.stId) { |
| 102 | + // Try to get contained events (sub-pathways and reactions) |
| 103 | + if (detailed.schemaClass === "Pathway" || detailed.schemaClass === "TopLevelPathway") { |
| 104 | + try { |
| 105 | + const containedEvents = await contentClient.get<Event[]>( |
| 106 | + `/data/pathway/${encodeURIComponent(detailed.stId)}/containedEvents` |
| 107 | + ); |
| 108 | + |
| 109 | + const subPathways = containedEvents.filter( |
| 110 | + (e) => e.schemaClass === "Pathway" |
| 111 | + ); |
| 112 | + const reactions = containedEvents.filter( |
| 113 | + (e) => e.schemaClass === "Reaction" || e.schemaClass === "BlackBoxEvent" |
| 114 | + ); |
| 115 | + |
| 116 | + if (subPathways.length > 0) { |
| 117 | + lines.push("### Sub-pathways"); |
| 118 | + const limit = detailLevel === "detailed" ? 15 : 5; |
| 119 | + subPathways.slice(0, limit).forEach((p) => { |
| 120 | + lines.push(`- **${p.displayName}** (\`${p.stId}\`)`); |
| 121 | + }); |
| 122 | + if (subPathways.length > limit) { |
| 123 | + lines.push(`- *...and ${subPathways.length - limit} more*`); |
| 124 | + } |
| 125 | + lines.push(""); |
| 126 | + } |
| 127 | + |
| 128 | + if (reactions.length > 0) { |
| 129 | + lines.push(`### Reactions`); |
| 130 | + lines.push(`This pathway involves **${reactions.length}** reactions.`); |
| 131 | + if (detailLevel === "detailed") { |
| 132 | + reactions.slice(0, 10).forEach((r) => { |
| 133 | + lines.push(`- ${r.displayName} (\`${r.stId}\`)`); |
| 134 | + }); |
| 135 | + if (reactions.length > 10) { |
| 136 | + lines.push(`- *...and ${reactions.length - 10} more*`); |
| 137 | + } |
| 138 | + } |
| 139 | + lines.push(""); |
| 140 | + } |
| 141 | + } catch { |
| 142 | + // Not a pathway or fetch failed — skip |
| 143 | + } |
| 144 | + } |
| 145 | + } |
| 146 | + |
| 147 | + // Step 6: Literature references |
| 148 | + if ( |
| 149 | + detailed?.literatureReference && |
| 150 | + detailed.literatureReference.length > 0 |
| 151 | + ) { |
| 152 | + const limit = detailLevel === "brief" ? 2 : detailLevel === "standard" ? 3 : 5; |
| 153 | + lines.push("### Literature References"); |
| 154 | + detailed.literatureReference.slice(0, limit).forEach((ref) => { |
| 155 | + if (ref.pubMedIdentifier) { |
| 156 | + lines.push( |
| 157 | + `- [${ref.displayName}](https://pubmed.ncbi.nlm.nih.gov/${ref.pubMedIdentifier})` |
| 158 | + ); |
| 159 | + } else { |
| 160 | + lines.push(`- ${ref.displayName}`); |
| 161 | + } |
| 162 | + }); |
| 163 | + if (detailed.literatureReference.length > limit) { |
| 164 | + lines.push( |
| 165 | + `- *...and ${detailed.literatureReference.length - limit} more references*` |
| 166 | + ); |
| 167 | + } |
| 168 | + lines.push(""); |
| 169 | + } |
| 170 | + |
| 171 | + // Step 7: Related results from search |
| 172 | + if (detailLevel === "detailed" && entries.length > 1) { |
| 173 | + lines.push("### Related Entries in Reactome"); |
| 174 | + entries.slice(1, 6).forEach((entry) => { |
| 175 | + const entryName = stripHtml(entry.name); |
| 176 | + lines.push(`- **${entryName}** (\`${entry.stId}\`) — ${entry.exactType}`); |
| 177 | + }); |
| 178 | + lines.push(""); |
| 179 | + } |
| 180 | + |
| 181 | + // Step 8: Helpful next steps |
| 182 | + lines.push("### Explore Further"); |
| 183 | + if (detailed?.stId) { |
| 184 | + lines.push(`- Use \`reactome_get_pathway\` with ID \`${detailed.stId}\` for full details`); |
| 185 | + if (detailed.hasDiagram) { |
| 186 | + lines.push(`- Use \`reactome_export_diagram\` with ID \`${detailed.stId}\` to get the pathway diagram`); |
| 187 | + } |
| 188 | + lines.push(`- Use \`reactome_participants\` with ID \`${detailed.stId}\` to see molecular participants`); |
| 189 | + } |
| 190 | + lines.push( |
| 191 | + `- Use \`reactome_search\` with query \`${query}\` to see all matching results` |
| 192 | + ); |
| 193 | + |
| 194 | + return lines.join("\n"); |
| 195 | +} |
| 196 | + |
| 197 | +export function registerAiExplainTools(server: McpServer) { |
| 198 | + server.tool( |
| 199 | + "reactome_ai_explain", |
| 200 | + "Get a clear, human-readable explanation of a biological concept, pathway, or entity using data from the Reactome knowledgebase. No API key required.", |
| 201 | + { |
| 202 | + query: z |
| 203 | + .string() |
| 204 | + .describe( |
| 205 | + 'The biological concept or entity to explain (e.g., "TP53", "apoptosis", "cell cycle", "BRCA1")' |
| 206 | + ), |
| 207 | + detail_level: z |
| 208 | + .enum(["brief", "standard", "detailed"]) |
| 209 | + .optional() |
| 210 | + .default("standard") |
| 211 | + .describe( |
| 212 | + "How detailed the explanation should be: brief (summary only), standard (with sub-pathways), detailed (comprehensive with reactions and references)" |
| 213 | + ), |
| 214 | + }, |
| 215 | + async ({ query, detail_level }) => { |
| 216 | + try { |
| 217 | + const explanation = await buildExplanation(query, detail_level); |
| 218 | + return { |
| 219 | + content: [{ type: "text", text: explanation }], |
| 220 | + }; |
| 221 | + } catch (error) { |
| 222 | + return { |
| 223 | + content: [ |
| 224 | + { |
| 225 | + type: "text", |
| 226 | + text: `## Error\n\nFailed to fetch data from Reactome: ${error instanceof Error ? error.message : String(error)}`, |
| 227 | + }, |
| 228 | + ], |
| 229 | + }; |
| 230 | + } |
| 231 | + } |
| 232 | + ); |
| 233 | +} |
0 commit comments