@sjcrh/proteinpaint-server 2.188.1 → 2.190.0

package/routes/profile.radarFacility2.js

@@ -1,148 +0,0 @@
-import { ProfileScoresPayload } from "#types/checkers";
-import { getData } from "../src/termdb.matrix.js";
-const api = {
-  endpoint: "termdb/profileRadarFacility2Scores",
-  methods: {
-    get: {
-      ...ProfileScoresPayload,
-      init
-    },
-    post: {
-      ...ProfileScoresPayload,
-      init
-    }
-  }
-};
-function init({ genomes }) {
-  return async (req, res) => {
-    try {
-      const g = genomes[req.query.genome];
-      if (!g) throw "invalid genome name";
-      const ds = g.datasets?.[req.query.dslabel];
-      const result = await getScores(req.query, ds);
-      res.send(result);
-    } catch (e) {
-      console.log(e);
-      res.send({ status: "error", error: e.message || e });
-    }
-  };
-}
-function derivePrefix(query) {
-  const firstScoreId = query.scoreTerms?.[0]?.score?.term?.id;
-  if (firstScoreId?.startsWith("F")) return "F";
-  if (firstScoreId?.startsWith("A")) return "A";
-  for (const entry of query.filter?.lst || []) {
-    const id = entry.tvs?.term?.id;
-    if (id?.startsWith("F")) return "F";
-    if (id?.startsWith("A")) return "A";
-  }
-  throw "cannot determine cohort prefix from scoreTerms or filter term IDs";
-}
-async function getScores(query, ds) {
-  const { activeCohort, clientAuthResult } = query.__protected__;
-  const prefix = derivePrefix(query);
-  const facilityTermId = `${prefix}UNIT`;
-  const facilityTW = { term: { id: facilityTermId }, q: {} };
-  const terms = [facilityTW];
-  for (const t of query.scoreTerms) {
-    terms.push(t.score);
-    if (t.maxScore?.term) terms.push(t.maxScore);
-  }
-  if (!query.filterByUserSites) {
-    query.__protected__.ignoredTermIds.push(facilityTermId);
-  }
-  const cohortAuth = clientAuthResult[activeCohort];
-  const isPublic = !cohortAuth?.role || cohortAuth.role === "public";
-  const isAdmin = cohortAuth?.role === "admin";
-  const userSites = cohortAuth?.sites;
-  const raw = await getData(
-    {
-      terms,
-      filter: query.filter,
-      __protected__: query.__protected__
-    },
-    ds
-  );
-  if (raw.error) throw raw.error;
-  const samples = Object.values(raw.samples);
-  let sites = samples.map((s) => {
-    const val = s[facilityTW.$id].value;
-    let label = facilityTW.term.values?.[val]?.label || val;
-    if (label.length > 50) label = label.slice(0, 47) + "...";
-    return { value: val, label };
-  });
-  if (userSites && !isAdmin) {
-    sites = sites.filter((s) => userSites.includes(s.value));
-  }
-  sites.sort((a, b) => a.label.localeCompare(b.label));
-  const eligibleSamples = userSites && query.filterByUserSites ? samples.filter((s) => userSites.includes(s[facilityTW.$id].value)) : samples;
-  const aggregateScore = {};
-  for (const d of query.scoreTerms) {
-    const score = computeMedianPercentage(d, eligibleSamples);
-    if (score !== null) aggregateScore[d.score.term.id] = score;
-  }
-  let targetSiteValue = null;
-  if (query.facilitySite) {
-    if (isAdmin) targetSiteValue = query.facilitySite;
-    else if (userSites?.includes(query.facilitySite)) targetSiteValue = query.facilitySite;
-  }
-  if (!targetSiteValue && sites.length > 0) targetSiteValue = sites[0].value;
-  let sampleData = void 0;
-  if (!isPublic && targetSiteValue) {
-    const sampleRow = samples.find((s) => s[facilityTW.$id].value == targetSiteValue);
-    if (sampleRow) {
-      const site = sites.find((s) => s.value == targetSiteValue) || {
-        value: targetSiteValue,
-        label: facilityTW.term.values?.[targetSiteValue]?.label || targetSiteValue
-      };
-      const singleSiteScore = {};
-      for (const d of query.scoreTerms) {
-        const percent = computeSinglePercentage(d, sampleRow);
-        if (percent !== null) singleSiteScore[d.score.term.id] = percent;
-      }
-      sampleData = { term2Score: singleSiteScore, site, sites, n: 1 };
-    }
-  }
-  return {
-    term2Score: aggregateScore,
-    sampleData,
-    // Public users: empty (chart should be unreachable anyway via isSupportedChartOverride)
-    sites: isPublic ? [] : sites,
-    n: eligibleSamples.length
-  };
-}
-function computeSinglePercentage(d, sample) {
-  const scoreValue = sample[d.score.$id]?.value;
-  if (scoreValue == null) return null;
-  let maxScoreValue = null;
-  if (typeof d.maxScore === "number") {
-    maxScoreValue = d.maxScore;
-  } else {
-    maxScoreValue = sample[d.maxScore.$id]?.value;
-  }
-  if (maxScoreValue == null || maxScoreValue === 0) return null;
-  return Math.round(scoreValue / maxScoreValue * 100);
-}
-function computeMedianPercentage(d, samples) {
-  const percentages = [];
-  for (const s of samples) {
-    const scoreValue = s[d.score.$id]?.value;
-    if (scoreValue == null) continue;
-    let maxScoreValue = null;
-    if (typeof d.maxScore === "number") {
-      maxScoreValue = d.maxScore;
-    } else {
-      maxScoreValue = s[d.maxScore.$id]?.value;
-    }
-    if (maxScoreValue == null || maxScoreValue === 0) continue;
-    percentages.push(scoreValue / maxScoreValue * 100);
-  }
-  if (percentages.length === 0) return null;
-  percentages.sort((a, b) => a - b);
-  const mid = Math.floor(percentages.length / 2);
-  const median = percentages.length % 2 !== 0 ? percentages[mid] : (percentages[mid - 1] + percentages[mid]) / 2;
-  return Math.round(median);
-}
-export {
-  api
-};

package/routes/sampledzimages.js

@@ -1,48 +0,0 @@
-import fs from "fs";
-import path from "path";
-import serverconfig from "#src/serverconfig.js";
-import { dzImagesPayload } from "#types/checkers";
-const api = {
-  endpoint: "sampledzimages",
-  methods: {
-    get: {
-      init,
-      ...dzImagesPayload
-    },
-    post: {
-      init,
-      ...dzImagesPayload
-    }
-  }
-};
-function init({ genomes }) {
-  return async (req, res) => {
-    const q = req.query;
-    try {
-      const g = genomes[q.genome];
-      if (!g) throw "invalid genome name";
-      const ds = g.datasets[q.dslabel];
-      if (!ds) throw "invalid dataset name";
-      const sampleId = q.sample_id || "";
-      const sampleDZImagesPath = path.join(
-        `${serverconfig.tpmasterdir}/${ds.queries.DZImages.imageBySampleFolder}`,
-        sampleId
-      );
-      const sampleDZImages = getDZImages(sampleDZImagesPath);
-      res.send(
-        { sampleDZImages }
-        /*satisfies DZImagesResponse*/
-      );
-    } catch (e) {
-      console.log(e);
-      res.status(404).send("Sample images not found");
-    }
-  };
-}
-function getDZImages(sampleImagesPath) {
-  const files = fs.readdirSync(sampleImagesPath);
-  return files.filter((file) => path.extname(file) === ".dzi");
-}
-export {
-  api
-};

package/routes/samplewsimages.js

@@ -1,60 +0,0 @@
-import { sampleWSImagesPayload } from "#types/checkers";
-const api = {
-  endpoint: "samplewsimages",
-  methods: {
-    get: {
-      ...sampleWSImagesPayload,
-      init
-    },
-    post: {
-      ...sampleWSImagesPayload,
-      init
-    }
-  }
-};
-function init({ genomes }) {
-  return async (req, res) => {
-    try {
-      const query = req.query;
-      const g = genomes[query.genome];
-      if (!g) throw "invalid genome name";
-      const ds = g.datasets[query.dslabel];
-      if (!ds) throw "invalid dataset name";
-      const sampleId = query.sample_id;
-      const wsimages = await ds.queries.WSImages.getWSImages(sampleId);
-      res.send({ sampleWSImages: wsimages });
-    } catch (e) {
-      console.log(e);
-      res.status(404).send("Sample images not found");
-    }
-  };
-}
-async function validate_query_getSampleWSImages(ds) {
-  if (!ds.queries?.WSImages) return;
-  validateQuery(ds);
-}
-function validateQuery(ds) {
-  if (typeof ds.queries.WSImages.getWSImages == "function") {
-    return;
-  }
-  ds.queries.WSImages.getWSImages = async (sampleName) => {
-    const sql = `SELECT wsimages.filename as filename, wsimages.metadata as metadata
-					 FROM wsimages
-							  INNER JOIN sampleidmap
-										 ON wsimages.sample = sampleidmap.id
-					 WHERE sampleidmap.name = ?`;
-    const rows = ds.cohort.db.connection.prepare(sql).all(sampleName);
-    const images = [];
-    for (const row of rows) {
-      images.push({
-        filename: row.filename,
-        metadata: row.metadata
-      });
-    }
-    return images;
-  };
-}
-export {
-  api,
-  validate_query_getSampleWSImages
-};

package/routes/snp.js

@@ -1,98 +0,0 @@
-import { snpPayload } from "#types/checkers";
-import * as utils from "#src/utils.js";
-const api = {
-  // route endpoint
-  // - no need for trailing slash
-  // - should be a noun (method is based on HTTP GET, POST, etc)
-  // - don't add 'Data' as response is assumed to be data
-  endpoint: "snp",
-  methods: {
-    get: {
-      ...snpPayload,
-      init
-    },
-    post: {
-      ...snpPayload,
-      init
-    }
-  }
-};
-function init({ genomes }) {
-  return async function handle_snp(req, res) {
-    try {
-      const q = req.query;
-      const n = q.genome;
-      if (!n) throw "no genome";
-      const results = await searchSNP(q, genomes[n]);
-      res.send({ results });
-    } catch (e) {
-      if (e.stack) console.log(e.stack);
-      return res.send({ error: e.message || e });
-    }
-  };
-}
-async function searchSNP(q, genome) {
-  if (!genome) throw "invalid genome";
-  if (!genome.snp) throw "snp is not configured for this genome";
-  const hits = [];
-  if (q.byCoord) {
-    if (genome.genomicNameRegexp.test(q.chr)) throw "invalid chr name";
-    if (!Array.isArray(q.ranges)) throw "ranges not an array";
-    for (const r of q.ranges) {
-      if (!Number.isInteger(r.start) || !Number.isInteger(r.stop) || r.start < 0 || r.stop < r.start)
-        throw "invalid start/stop";
-      if (r.stop - r.start >= 100) {
-        throw "range too big";
-      }
-      const snps = await utils.query_bigbed_by_coord(genome.snp.bigbedfile, q.chr, r.start, r.stop);
-      for (const snp of snps) {
-        const hit = snp2hit(snp);
-        if (q.alleleLst) {
-          let missing = false;
-          for (const i of q.alleleLst) {
-            if (i && !hit.alleles.includes(i)) {
-              missing = true;
-              break;
-            }
-          }
-          if (missing) continue;
-        }
-        hits.push(hit);
-      }
-    }
-  } else if (q.byName) {
-    if (!Array.isArray(q.lst)) throw ".lst[] missing";
-    for (const n of q.lst) {
-      if (genome.genomicNameRegexp.test(n)) continue;
-      const snps = await utils.query_bigbed_by_name(genome.snp.bigbedfile, n);
-      for (const snp of snps) {
-        const hit = snp2hit(snp);
-        hits.push(hit);
-      }
-    }
-  } else {
-    throw "unknown query method";
-  }
-  return hits;
-}
-function snp2hit(snp) {
-  const fields = snp.split("	");
-  const ref = fields[4];
-  const alts = fields[6].split(",").filter(Boolean);
-  const observed = ref + "/" + alts.join("/");
-  const hit = {
-    chrom: fields[0],
-    chromStart: Number(fields[1]),
-    chromEnd: Number(fields[2]),
-    name: fields[3],
-    observed,
-    ref,
-    alt: alts,
-    alleles: [ref, ...alts]
-  };
-  return hit;
-}
-export {
-  api,
-  searchSNP
-};

package/routes/termdb.chat2.js

@@ -1,217 +0,0 @@
-import { ChatPayload } from "#types/checkers";
-import { classifyQuery } from "./chat/classify1.ts";
-import { classifyNotPlot } from "./chat/classify2.ts";
-import { classifyPlotType } from "./chat/plot.ts";
-import { extract_DE_search_terms_from_query } from "./chat/DEagent.ts";
-import { determineAmbiguousGenePrompt } from "./chat/ambiguousgeneagent.ts";
-import { extract_summary_terms } from "./chat/summaryagent.ts";
-import { extract_matrix_search_terms_from_query } from "./chat/matrixagent.ts";
-import { extract_samplescatter_terms_from_query } from "./chat/samplescatteragent.ts";
-import { extract_hiercluster_terms_from_query } from "./chat/hierclusteragent.ts";
-import { classifyGeneDataType } from "./chat/genedatatypeagent.ts";
-import { extractGenesFromPrompt, parse_dataset_db, parse_geneset_db, getGenesetNames } from "./chat/utils.ts";
-import serverconfig from "../src/serverconfig.js";
-import { mayLog } from "#src/helpers.ts";
-import { formatElapsedTime } from "#shared";
-import path from "path";
-import fs from "fs";
-import { readJSONFile } from "./chat/utils.ts";
-const api = {
-  endpoint: "termdb/chat2",
-  methods: {
-    get: {
-      ...ChatPayload,
-      init
-    },
-    post: {
-      ...ChatPayload,
-      init
-    }
-  }
-};
-function init({ genomes }) {
-  return async (req, res) => {
-    const q = req.query;
-    try {
-      const g = genomes[q.genome];
-      if (!g) throw "invalid genome";
-      const ds = g.datasets?.[q.dslabel];
-      if (!ds) throw "invalid dslabel";
-      const aiFilesDir = serverconfig.binpath + "/../../dataset/ai/" + q.dslabel;
-      let agentFiles = [];
-      try {
-        agentFiles = fs.readdirSync(aiFilesDir).filter((file) => file.endsWith(".json"));
-      } catch (err) {
-        if (err.code === "ENOENT") throw new Error(`Directory not found: ${aiFilesDir}`);
-        if (err.code === "ENOTDIR") throw new Error(`Path is not a directory: ${aiFilesDir}`);
-        throw err;
-      }
-      const llm = serverconfig.llm;
-      if (!llm) throw "serverconfig.llm is not configured";
-      if (llm.provider !== "SJ" && llm.provider !== "ollama") {
-        throw "llm.provider must be 'SJ' or 'ollama'";
-      }
-      const dataset_db = serverconfig.tpmasterdir + "/" + ds.cohort.db.file;
-      const genedb = serverconfig.tpmasterdir + "/" + g.genedb.dbfile;
-      const testing = false;
-      const genesetNames = getGenesetNames(g);
-      const ai_output_json = await run_chat_pipeline(
-        q.prompt,
-        llm,
-        testing,
-        dataset_db,
-        genedb,
-        ds,
-        genesetNames,
-        agentFiles,
-        aiFilesDir
-      );
-      res.send(ai_output_json);
-    } catch (e) {
-      if (e.stack) mayLog(e.stack);
-      res.send({ error: e?.message || e });
-    }
-  };
-}
-async function run_chat_pipeline(user_prompt, llm, testing, dataset_db, genedb, ds, genesetNames = [], agentFiles, aiFilesDir) {
-  if (!fs.existsSync(path.join(aiFilesDir, "main.json")))
-    throw "Main data file is not specified for dataset:" + ds.label;
-  const dataset_json = await readJSONFile(path.join(aiFilesDir, "main.json"));
-  const time1 = (/* @__PURE__ */ new Date()).valueOf();
-  const class_response = await classifyQuery(user_prompt, llm);
-  let ai_output_json;
-  mayLog("Time taken for classification:", formatElapsedTime(Date.now() - time1));
-  if (class_response.type == "notplot") {
-    const time2 = (/* @__PURE__ */ new Date()).valueOf();
-    const notPlotResult = await classifyNotPlot(user_prompt, llm, agentFiles, aiFilesDir);
-    mayLog("Time taken for classify2:", formatElapsedTime(Date.now() - time2));
-    if (notPlotResult.type == "html") {
-      ai_output_json = notPlotResult;
-    } else {
-      ai_output_json = {
-        type: "text",
-        text: "Your query does not appear to be related to the available data visualizations. Please try rephrasing your question."
-      };
-    }
-  } else if (class_response.type == "plot") {
-    let geneFeatures = [];
-    const genes_list = await parse_geneset_db(genedb);
-    const relevant_genes = extractGenesFromPrompt(user_prompt, genes_list);
-    if (relevant_genes.length > 0) {
-      const AmbiguousGeneMessage = determineAmbiguousGenePrompt(user_prompt, relevant_genes, dataset_json);
-      if (AmbiguousGeneMessage.length > 0) {
-        return {
-          type: "text",
-          text: AmbiguousGeneMessage
-        };
-      }
-      const geneDataTypeMessage = await classifyGeneDataType(
-        user_prompt,
-        llm,
-        relevant_genes,
-        dataset_json
-      );
-      if (typeof geneDataTypeMessage === "string" || geneDataTypeMessage instanceof String) {
-        if (geneDataTypeMessage.length > 0) {
-          return {
-            type: "text",
-            text: geneDataTypeMessage
-          };
-        } else {
-          throw "classifyGeneDataType agent returned an empty string, which is unexpected.";
-        }
-      } else if (Array.isArray(geneDataTypeMessage)) {
-        geneFeatures = geneDataTypeMessage;
-      } else {
-        throw "geneDataTypeMessage has unknown data type returned from classifyGeneDataType agent";
-      }
-    }
-    const classResult = await classifyPlotType(user_prompt, llm);
-    const dataset_db_output = await parse_dataset_db(dataset_db);
-    if (classResult == "summary") {
-      const time12 = (/* @__PURE__ */ new Date()).valueOf();
-      ai_output_json = await extract_summary_terms(
-        user_prompt,
-        llm,
-        dataset_db_output,
-        dataset_json,
-        ds,
-        testing,
-        genesetNames,
-        geneFeatures,
-        aiFilesDir
-      );
-      mayLog("Time taken for summary agent:", formatElapsedTime(Date.now() - time12));
-    } else if (classResult == "dge") {
-      const time12 = (/* @__PURE__ */ new Date()).valueOf();
-      ai_output_json = await extract_DE_search_terms_from_query(
-        user_prompt,
-        llm,
-        dataset_db_output,
-        dataset_json,
-        ds,
-        testing,
-        aiFilesDir
-      );
-      mayLog("Time taken for DE agent:", formatElapsedTime(Date.now() - time12));
-    } else if (classResult == "survival") {
-      ai_output_json = { type: "text", text: "survival agent has not been implemented yet" };
-    } else if (classResult == "matrix") {
-      const time12 = (/* @__PURE__ */ new Date()).valueOf();
-      ai_output_json = await extract_matrix_search_terms_from_query(
-        user_prompt,
-        llm,
-        dataset_db_output,
-        dataset_json,
-        ds,
-        testing,
-        genesetNames,
-        geneFeatures,
-        aiFilesDir
-      );
-      mayLog("Time taken for matrix agent:", formatElapsedTime(Date.now() - time12));
-    } else if (classResult == "samplescatter") {
-      const time12 = (/* @__PURE__ */ new Date()).valueOf();
-      ai_output_json = await extract_samplescatter_terms_from_query(
-        user_prompt,
-        llm,
-        dataset_db_output,
-        dataset_json,
-        ds,
-        testing,
-        genesetNames,
-        geneFeatures,
-        aiFilesDir
-      );
-      mayLog("Time taken for sampleScatter agent:", formatElapsedTime(Date.now() - time12));
-    } else if (classResult == "hiercluster") {
-      const time12 = (/* @__PURE__ */ new Date()).valueOf();
-      ai_output_json = await extract_hiercluster_terms_from_query(
-        user_prompt,
-        llm,
-        dataset_db_output,
-        dataset_json,
-        ds,
-        testing,
-        genesetNames,
-        geneFeatures,
-        aiFilesDir
-      );
-      mayLog("Time taken for hierCluster agent:", formatElapsedTime(Date.now() - time12));
-    } else if (classResult == "lollipop") {
-      ai_output_json = {
-        type: "text",
-        text: "This is a gene mutation prompt. But, lollipop agent has not been implemented yet"
-      };
-    } else {
-      ai_output_json = { type: "text", text: "Unknown classification value" };
-    }
-  } else {
-    ai_output_json = { type: "text", text: "Unknown classification type" };
-  }
-  return ai_output_json;
-}
-export {
-  api,
-  run_chat_pipeline
-};