@sjcrh/proteinpaint-server 2.190.2 → 2.191.1

package/routes/genesetEnrichment.js

@@ -1,226 +0,0 @@
-import fs from "fs";
-import path from "path";
-import serverconfig from "#src/serverconfig.js";
-import { run_python } from "@sjcrh/proteinpaint-python";
-import { run_rust } from "@sjcrh/proteinpaint-rust";
-import { mayLog } from "#src/helpers.ts";
-import { formatElapsedTime } from "#shared";
-import { getDeCacheResult } from "../routes/termdb.DE.ts";
-import { getDmCacheResult } from "../routes/termdb.diffMeth.ts";
-import { cacheOrRecompute } from "#src/utils/cacheOrRecompute.ts";
-import { get_ds_tdb } from "#src/termdb.js";
-function init({ genomes }) {
-  return async (req, res) => {
-    try {
-      const q = req.query;
-      const results = await run_genesetEnrichment_analysis(q, genomes);
-      if (!q.geneset_name) {
-        if (typeof results != "object") throw new Error("gsea result is not object");
-        res.send(results);
-        return;
-      }
-      if (typeof results === "object" && results.error) {
-        res.send({ status: "error", error: results.error });
-        return;
-      }
-      if (typeof results != "string") throw new Error("gsea result is not string");
-      res.sendFile(results, (err) => {
-        fs.unlink(results, () => {
-        });
-        if (err) {
-          res.status(404).send("Image not found");
-        }
-      });
-    } catch (e) {
-      res.status(e.status || 500).send({ status: "error", error: e.message || e, code: e.code });
-      if (e.stack) console.log(e.stack);
-    }
-  };
-}
-async function run_genesetEnrichment_analysis(q, genomes) {
-  if (!genomes[q.genome].termdbs) throw new Error("termdb database is not available for " + q.genome);
-  if (q.fetchDE) {
-    const { genes, fold_change } = await resolveGseaGenesAndFoldChange({ q, genomes });
-    return { data: { genes, fold_change } };
-  }
-  if (q.method == "blitzgsea") {
-    if (q.geneset_name) return await computeGseaImage({ q, genomes });
-    return await computeGseaInitial({ q, genomes });
-  }
-  if (q.method == "cerno") {
-    const { genes, fold_change } = await resolveGseaGenesAndFoldChange({ q, genomes });
-    const genesetenrichment_input = {
-      genes,
-      fold_change,
-      db: genomes[q.genome].termdbs.msigdb.cohort.db.connection.name,
-      geneset_group: q.geneSetGroup,
-      genedb: path.join(serverconfig.tpmasterdir, genomes[q.genome].genedb.dbfile),
-      filter_non_coding_genes: q.filter_non_coding_genes
-    };
-    const time1 = (/* @__PURE__ */ new Date()).valueOf();
-    const gsea_output = JSON.parse(await run_rust("cerno", JSON.stringify(genesetenrichment_input)));
-    mayLog("Time taken to run CERNO:", formatElapsedTime(Date.now() - time1));
-    return gsea_output;
-  }
-  throw new Error("Unknown method:" + q.method);
-}
-function gseaKeyInputs(q, genes, fold_change) {
-  return {
-    genes,
-    fold_change,
-    geneSetGroup: q.geneSetGroup,
-    num_permutations: q.num_permutations,
-    filter_non_coding_genes: q.filter_non_coding_genes
-  };
-}
-async function getGseaCacheResult({
-  q,
-  genomes
-}) {
-  const { genes, fold_change } = await resolveGseaGenesAndFoldChange({ q, genomes });
-  const cacheArg = gseaKeyInputs(q, genes, fold_change);
-  const { result, cacheId } = await cacheOrRecompute({
-    computeArgument: cacheArg,
-    cacheSubdir: "gsea",
-    computeFresh: async () => {
-      const { table, pickleB64 } = await runGseaPythonForTable({ q, genomes, cacheArg });
-      const cacheResult = { table, pickleB64 };
-      return cacheResult;
-    }
-  });
-  return { result, cacheId };
-}
-async function computeGseaInitial({
-  q,
-  genomes
-}) {
-  const { result } = await getGseaCacheResult({ q, genomes });
-  return result.table;
-}
-const pendingImageRequests = /* @__PURE__ */ new Map();
-async function computeGseaImage({
-  q,
-  genomes
-}) {
-  const { result, cacheId } = await getGseaCacheResult({ q, genomes });
-  const dedupKey = `${cacheId}:${q.geneset_name}`;
-  const inFlight = pendingImageRequests.get(dedupKey);
-  if (inFlight) return inFlight;
-  const work = runGseaPythonForImage({
-    q,
-    genomes,
-    pickleB64: result.pickleB64
-  }).finally(() => pendingImageRequests.delete(dedupKey));
-  pendingImageRequests.set(dedupKey, work);
-  return work;
-}
-async function runGseaPythonForTable({
-  q,
-  genomes,
-  cacheArg
-}) {
-  const pyInput = buildPyInput({ ...q, geneset_name: void 0 }, genomes, cacheArg);
-  const time1 = (/* @__PURE__ */ new Date()).valueOf();
-  const gsea_output = await run_python("gsea.py", "/" + JSON.stringify(pyInput));
-  mayLog("Time taken to run blitzgsea:", formatElapsedTime(Date.now() - time1));
-  for (const line of gsea_output.split("\n")) {
-    if (line.startsWith("result: ")) {
-      const parsed = JSON.parse(line.replace("result: ", ""));
-      if (parsed?.error) throw new Error(parsed.error);
-      const { pickle_b64, ...table } = parsed;
-      if (!pickle_b64) throw new Error("gsea.py result missing pickle_b64");
-      return { table, pickleB64: pickle_b64 };
-    }
-    mayLog(line);
-  }
-  throw new Error("gsea.py did not emit a result line on the initial path");
-}
-async function runGseaPythonForImage({
-  q,
-  genomes,
-  pickleB64
-}) {
-  const { genes, fold_change } = await resolveGseaGenesAndFoldChange({ q, genomes });
-  const cacheArg = gseaKeyInputs(q, genes, fold_change);
-  const pyInput = buildPyInput(q, genomes, cacheArg, pickleB64);
-  const time1 = (/* @__PURE__ */ new Date()).valueOf();
-  const gsea_output = await run_python("gsea.py", "/" + JSON.stringify(pyInput));
-  mayLog("Time taken to render gsea image:", formatElapsedTime(Date.now() - time1));
-  for (const line of gsea_output.split("\n")) {
-    if (line.startsWith("image: ")) {
-      const parsed = JSON.parse(line.replace("image: ", ""));
-      return path.join(serverconfig.cachedir, "gsea", parsed.image_file);
-    }
-    if (line.startsWith("result: ")) {
-      const parsed = JSON.parse(line.replace("result: ", ""));
-      if (parsed?.error) return { error: parsed.error };
-    }
-    mayLog(line);
-  }
-  throw new Error("gsea.py did not emit an image line on the detail path");
-}
-function buildPyInput(q, genomes, cacheArg, pickleB64) {
-  return {
-    genes: cacheArg.genes,
-    fold_change: cacheArg.fold_change,
-    db: genomes[q.genome].termdbs.msigdb.cohort.db.connection.name,
-    geneset_group: cacheArg.geneSetGroup,
-    genedb: path.join(serverconfig.tpmasterdir, genomes[q.genome].genedb.dbfile),
-    filter_non_coding_genes: cacheArg.filter_non_coding_genes,
-    cachedir: path.join(serverconfig.cachedir, "gsea"),
-    geneset_name: q.geneset_name,
-    num_permutations: cacheArg.num_permutations,
-    ...pickleB64 ? { pickle_b64: pickleB64 } : {}
-  };
-}
-async function resolveGseaGenesAndFoldChange({
-  q,
-  genomes
-}) {
-  if (q.cacheId) {
-    if (!q.daRequest) throw new Error("daCacheMissing");
-    const kind = q.daRequest.kind;
-    if (kind !== "DE" && kind !== "DM") throw new Error('daRequest.kind must be "DE" or "DM"');
-    if (kind === "DE") {
-      const { result: result2, cacheId: cacheId2 } = await getDeCacheResult(q.daRequest, genomes);
-      if (cacheId2 !== q.cacheId) throw new Error("cacheId does not match daRequest");
-      return {
-        genes: result2.geneRows.map((g) => g.gene_name),
-        fold_change: result2.geneRows.map((g) => g.fold_change)
-      };
-    }
-    const { result, cacheId } = await getDmCacheResult(q.daRequest, genomes);
-    if (cacheId !== q.cacheId) throw new Error("cacheId does not match daRequest");
-    return {
-      genes: result.promoterRows.map((p) => p.gene_name),
-      fold_change: result.promoterRows.map((p) => p.fold_change)
-    };
-  }
-  if (q.dapParams) {
-    const genome = genomes[q.genome];
-    if (!genome) throw new Error("invalid genome");
-    const [ds] = get_ds_tdb(genome, q);
-    const { organism, assay, cohort } = q.dapParams;
-    const cohortConfig = ds.queries?.proteome?.organisms?.[organism]?.assays?.[assay]?.cohorts?.[cohort];
-    if (!cohortConfig?.DAPfile) throw new Error("DAP file not configured for this cohort");
-    const filePath = path.join(serverconfig.tpmasterdir, cohortConfig.DAPfile);
-    const content = await fs.promises.readFile(filePath, "utf8");
-    const lines = content.trim().split("\n");
-    const genes = [];
-    const fold_change = [];
-    for (let i = 1; i < lines.length; i++) {
-      const parts = lines[i].split("	");
-      if (parts.length < 4) continue;
-      const fc = Number(parts[2]);
-      if (!Number.isFinite(fc)) continue;
-      genes.push(parts[1]);
-      fold_change.push(fc);
-    }
-    return { genes, fold_change };
-  }
-  if (!q.genes || !q.fold_change) throw new Error("requires genes and fold_change when cacheId is absent");
-  return { genes: q.genes, fold_change: q.fold_change };
-}
-export {
-  init
-};