@sjcrh/proteinpaint-server 2.118.2-0 → 2.118.2

package/dataset/termdb.test.js

@@ -192,7 +192,6 @@ function termdb_test_default() {
       },
       geneExpression: {
         src: "native",
-        hdf5File: true,
         file: "files/hg38/TermdbTest/TermdbTest.fpkm.matrix.h5"
       },
       topVariablyExpressedGenes: {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.118.2-0",
+  "version": "2.118.2",
   "type": "module",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "src/app.js",
@@ -65,11 +65,11 @@
     "typescript": "^5.6.3"
   },
   "dependencies": {
-    "@sjcrh/augen": "2.118.2-0",
+    "@sjcrh/augen": "2.118.2-1",
     "@sjcrh/proteinpaint-python": "2.118.0",
     "@sjcrh/proteinpaint-rust": "2.117.0",
     "@sjcrh/proteinpaint-shared": "2.118.0",
-    "@sjcrh/proteinpaint-types": "2.118.1-1",
+    "@sjcrh/proteinpaint-types": "2.118.2",
     "@types/express": "^5.0.0",
     "@types/express-session": "^1.18.1",
     "better-sqlite3": "^9.4.1",

package/routes/genesetEnrichment.js

@@ -3,6 +3,7 @@ import fs from "fs";
 import path from "path";
 import serverconfig from "#src/serverconfig.js";
 import { run_python } from "@sjcrh/proteinpaint-python";
+import { mayLog } from "#src/helpers.ts";
 const api = {
   endpoint: "genesetEnrichment",
   methods: {
@@ -22,24 +23,24 @@ function init({ genomes }) {
       const q = req.query;
       const results = await run_genesetEnrichment_analysis(q, genomes);
       if (!q.geneset_name) {
-        if (typeof results != "string")
-          res.send(results);
-        else
-          throw `invalid results type when !req.query.geneset_name`;
-      } else if (typeof results == "string") {
-        res.sendFile(results, (err) => {
-          fs.unlink(results, (del_err) => {
-            if (del_err) {
-              console.error("Error deleting file " + results + ":", del_err);
-            }
-          });
-          if (err) {
-            res.status(404).send("Image not found");
-          }
-        });
+        if (typeof results != "object")
+          throw "gsea result is not object";
+        res.send(results);
+        return;
       }
+      if (typeof results != "string")
+        throw "gsea result is not string";
+      res.sendFile(results, (err) => {
+        fs.unlink(results, () => {
+        });
+        if (err) {
+          res.status(404).send("Image not found");
+        }
+      });
     } catch (e) {
       res.send({ status: "error", error: e.message || e });
+      if (e.stack)
+        console.log(e.stack);
     }
   };
 }
@@ -71,17 +72,14 @@ async function run_genesetEnrichment_analysis(q, genomes) {
       result = JSON.parse(line.replace("image: ", ""));
       image_found = true;
     } else {
-      console.log(line);
+      mayLog(line);
     }
   }
-  if (data_found) {
+  if (data_found)
     return result;
-  } else if (image_found) {
-    const imagePath = path.join(serverconfig.cachedir, result.image_file);
-    return imagePath;
-  } else {
-    throw ``;
-  }
+  if (image_found)
+    return path.join(serverconfig.cachedir, result.image_file);
+  throw "data or image not found in gsea output; this should not happen";
 }
 export {
   api

package/routes/termdb.cluster.js

@@ -7,7 +7,6 @@ import serverconfig from "#src/serverconfig.js";
 import { gdc_validate_query_geneExpression } from "#src/mds3.gdc.js";
 import { mayLimitSamples } from "#src/mds3.filter.js";
 import { clusterMethodLst, distanceMethodLst } from "#shared/clustering.js";
-import { getResult as getResultGene } from "#src/gene.js";
 import { TermTypes, NUMERIC_DICTIONARY_TERM } from "#shared/terms.js";
 import { getData } from "#src/termdb.matrix.js";
 import { termType2label } from "#shared/terms.js";
@@ -220,7 +219,7 @@ async function validate_query_geneExpression(ds, genome) {
     return;
   }
   if (q.src == "native") {
-    await validateNative(q, ds, genome);
+    await validateNative(q, ds);
     return;
   }
   throw "unknown queries.geneExpression.src";
@@ -256,183 +255,89 @@ async function queryGeneExpression(hdf5_file, geneNames) {
     throw error;
   }
 }
-async function validateNative(q, ds, genome) {
-  if (q.hdf5File === true) {
-    q.file = path.join(serverconfig.tpmasterdir, q.file);
-    q.samples = [];
-    await utils.file_is_readable(q.file);
-    try {
-      const vr = await validateHDF5File(q.file);
-      if (vr.status !== "success")
-        throw vr.message;
-      if (!vr.sampleNames?.length)
-        throw "HDF5 file has no samples";
-      for (const sn of vr.sampleNames) {
-        const si = ds.cohort.termdb.q.sampleName2id(sn);
-        if (si == void 0)
-          throw "unknown sample from HDF5: " + sn;
-        q.samples.push(si);
-      }
-      console.log(`${ds.label}: HDF5 file validated. Format: ${vr.format}, Samples:`, vr.sampleNames.length);
-    } catch (error) {
-      throw `${ds.label}: Failed to validate HDF5 file: ${error}`;
+async function validateNative(q, ds) {
+  q.file = path.join(serverconfig.tpmasterdir, q.file);
+  q.samples = [];
+  await utils.file_is_readable(q.file);
+  try {
+    const vr = await validateHDF5File(q.file);
+    if (vr.status !== "success")
+      throw vr.message;
+    if (!vr.sampleNames?.length)
+      throw "HDF5 file has no samples";
+    for (const sn of vr.sampleNames) {
+      const si = ds.cohort.termdb.q.sampleName2id(sn);
+      if (si == void 0)
+        throw "unknown sample from HDF5: " + sn;
+      q.samples.push(si);
     }
-    q.get = async (param) => {
-      const limitSamples = await mayLimitSamples(param, q.samples, ds);
-      if (limitSamples?.size == 0) {
-        return { term2sample2value: /* @__PURE__ */ new Map(), byTermId: {}, bySampleId: {} };
-      }
-      const bySampleId = {};
-      const samples = q.samples || [];
-      if (limitSamples) {
-        for (const sid of limitSamples) {
-          bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
-        }
-      } else {
-        for (const sid of samples) {
-          bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
-        }
-      }
-      const term2sample2value = /* @__PURE__ */ new Map();
-      const byTermId = {};
-      const geneNames = [];
-      for (const geneTerm of param.terms) {
-        if (geneTerm.gene) {
-          geneNames.push(geneTerm.gene);
-        }
-      }
-      if (geneNames.length === 0) {
-        console.log("No genes to query");
-        return { term2sample2value, byTermId };
-      }
-      const time1 = Date.now();
-      try {
-        const geneData = JSON.parse(await queryGeneExpression(q.file, geneNames));
-        mayLog("Time taken to run gene query:", formatElapsedTime(Date.now() - time1));
-        const genesData = geneData.genes || { [geneNames[0]]: geneData };
-        for (const geneTerm of param.terms) {
-          if (!geneTerm.gene)
-            continue;
-          const geneResult = genesData[geneTerm.gene];
-          if (!geneResult) {
-            console.warn(`No data found for gene ${geneTerm.gene} in the response`);
-            continue;
-          }
-          const samplesData = geneResult.samples || {};
-          const s2v = {};
-          for (const [sampleName, value] of Object.entries(samplesData)) {
-            const sampleId = ds.cohort.termdb.q.sampleName2id(sampleName);
-            if (!sampleId)
-              continue;
-            if (limitSamples && !limitSamples.has(sampleId))
-              continue;
-            s2v[sampleId] = value;
-          }
-          if (Object.keys(s2v).length) {
-            term2sample2value.set(geneTerm.gene, s2v);
-          }
-        }
-      } catch (error) {
-        console.error(`Error processing batch gene query:`, error);
-      }
-      if (term2sample2value.size == 0) {
-        throw "No data available for the input " + param.terms?.map((g) => g.gene).join(", ");
-      }
-      return { term2sample2value, byTermId, bySampleId };
-    };
-  } else {
-    if (!q.file.startsWith(serverconfig.tpmasterdir)) {
-      q.file = path.join(serverconfig.tpmasterdir, q.file);
+    console.log(`${ds.label}: HDF5 file validated. Format: ${vr.format}, Samples:`, vr.sampleNames.length);
+  } catch (error) {
+    throw `${ds.label}: Failed to validate HDF5 file: ${error}`;
+  }
+  q.get = async (param) => {
+    const limitSamples = await mayLimitSamples(param, q.samples, ds);
+    if (limitSamples?.size == 0) {
+      return { term2sample2value: /* @__PURE__ */ new Map(), byTermId: {}, bySampleId: {} };
     }
-    if (!q.samples)
-      q.samples = [];
-    await utils.validate_tabixfile(q.file);
-    q.nochr = await utils.tabix_is_nochr(q.file, null, genome);
-    q.samples = [];
-    {
-      const lines = await utils.get_header_tabix(q.file);
-      if (!lines[0])
-        throw "Header line missing from " + q.file;
-      const l = lines[0].split("	");
-      if (l.slice(0, 4).join("	") != "#chr	start	stop	gene") {
-        throw "Header line has wrong content for columns 1-4";
+    const bySampleId = {};
+    const samples = q.samples || [];
+    if (limitSamples) {
+      for (const sid of limitSamples) {
+        bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
       }
-      for (let i = 4; i < l.length; i++) {
-        const id = ds.cohort.termdb.q.sampleName2id(l[i]);
-        if (id == void 0) {
-          throw "queries.geneExpression: unknown sample from header: " + l[i];
-        }
-        q.samples.push(id);
+    } else {
+      for (const sid of samples) {
+        bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
       }
     }
-    q.get = async (param) => {
-      const limitSamples = await mayLimitSamples(param, q.samples, ds);
-      if (limitSamples?.size == 0) {
-        return { term2sample2value: /* @__PURE__ */ new Map(), byTermId: {}, bySampleId: {} };
+    const term2sample2value = /* @__PURE__ */ new Map();
+    const byTermId = {};
+    const geneNames = [];
+    for (const geneTerm of param.terms) {
+      if (geneTerm.gene) {
+        geneNames.push(geneTerm.gene);
       }
-      const bySampleId = {};
-      const samples = q.samples || [];
-      if (limitSamples) {
-        for (const sid of limitSamples) {
-          bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
-        }
-      } else {
-        for (const sid of samples) {
-          bySampleId[sid] = { label: ds.cohort.termdb.q.id2sampleName(sid) };
-        }
-      }
-      const term2sample2value = /* @__PURE__ */ new Map();
+    }
+    if (geneNames.length === 0) {
+      console.log("No genes to query");
+      return { term2sample2value, byTermId };
+    }
+    const time1 = Date.now();
+    try {
+      const geneData = JSON.parse(await queryGeneExpression(q.file, geneNames));
+      mayLog("Time taken to run gene query:", formatElapsedTime(Date.now() - time1));
+      const genesData = geneData.genes || { [geneNames[0]]: geneData };
       for (const geneTerm of param.terms) {
         if (!geneTerm.gene)
           continue;
-        if (!geneTerm.chr || !Number.isInteger(geneTerm.start) || !Number.isInteger(geneTerm.stop)) {
-          const re = getResultGene(genome, { input: geneTerm.gene, deep: 1 });
-          if (!re.gmlst || re.gmlst.length == 0) {
-            console.warn("Unknown gene:" + geneTerm.gene);
-            continue;
-          }
-          const i = re.gmlst.find((i2) => i2.isdefault) || re.gmlst[0];
-          geneTerm.start = i.start;
-          geneTerm.stop = i.stop;
-          geneTerm.chr = i.chr;
+        const geneResult = genesData[geneTerm.gene];
+        if (!geneResult) {
+          console.warn(`No data found for gene ${geneTerm.gene} in the response`);
+          continue;
         }
+        const samplesData = geneResult.samples || {};
         const s2v = {};
-        if (!geneTerm.chr || !Number.isInteger(geneTerm.start) || !Number.isInteger(geneTerm.stop)) {
-          throw "Missing chr/start/stop";
+        for (const [sampleName, value] of Object.entries(samplesData)) {
+          const sampleId = ds.cohort.termdb.q.sampleName2id(sampleName);
+          if (!sampleId)
+            continue;
+          if (limitSamples && !limitSamples.has(sampleId))
+            continue;
+          s2v[sampleId] = value;
         }
-        await utils.get_lines_bigfile({
-          args: [
-            q.file,
-            (q.nochr ? geneTerm.chr.replace("chr", "") : geneTerm.chr) + ":" + geneTerm.start + "-" + geneTerm.stop
-          ],
-          callback: (line) => {
-            const l = line.split("	");
-            if (l[3].toLowerCase() != geneTerm.gene.toLowerCase())
-              return;
-            for (let i = 4; i < l.length; i++) {
-              const sampleId = samples[i - 4];
-              if (limitSamples && !limitSamples.has(sampleId))
-                continue;
-              if (!l[i])
-                continue;
-              const v = Number(l[i]);
-              if (Number.isNaN(v))
-                throw "Expression value not number";
-              s2v[sampleId] = v;
-            }
-          }
-        });
         if (Object.keys(s2v).length) {
           term2sample2value.set(geneTerm.gene, s2v);
         }
       }
-      const byTermId = {};
-      if (term2sample2value.size == 0) {
-        throw "No data available for the input " + param.terms?.map((g) => g.gene).join(", ");
-      }
-      return { term2sample2value, byTermId, bySampleId };
-    };
-  }
+    } catch (error) {
+      console.error(`Error processing batch gene query:`, error);
+    }
+    if (term2sample2value.size == 0) {
+      throw "No data available for the input " + param.terms?.map((g) => g.gene).join(", ");
+    }
+    return { term2sample2value, byTermId, bySampleId };
+  };
 }
 export {
   api,