@sjcrh/proteinpaint-server 2.120.0 → 2.120.2-0

package/dataset/termdb.test.js

@@ -154,12 +154,6 @@ function termdb_test_default() {
       cnv: {
         file: "files/hg38/TermdbTest/TermdbTest_CNV_gene.gz"
       },
-      /*
-      on the fly cnv calls from gene body probe signals are no longer used
-      probe2cnv:{
-      	file: 'files/hg19/pnet/PNET.probesignals.gz'
-      }
-      */
       singleSampleMutation: {
         src: "native",
         sample_id_key: "sample_id",
@@ -197,6 +191,10 @@ function termdb_test_default() {
       topVariablyExpressedGenes: {
         src: "native"
       },
+      rnaseqGeneCount: {
+        storage_type: "HDF5",
+        file: "files/hg38/TermdbTest/TermdbTest.geneCounts.h5"
+      },
       WSImages: {
         type: "H&E",
         imageBySampleFolder: "files/hg38/TermdbTest/wsimages"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.120.0",
+  "version": "2.120.2-0",
   "type": "module",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "src/app.js",
@@ -65,10 +65,10 @@
     "typescript": "^5.6.3"
   },
   "dependencies": {
-    "@sjcrh/augen": "2.119.0",
+    "@sjcrh/augen": "2.120.2-0",
     "@sjcrh/proteinpaint-python": "2.118.0",
-    "@sjcrh/proteinpaint-rust": "2.120.0",
-    "@sjcrh/proteinpaint-shared": "2.119.0",
+    "@sjcrh/proteinpaint-rust": "2.120.2-0",
+    "@sjcrh/proteinpaint-shared": "2.120.1",
     "@sjcrh/proteinpaint-types": "2.119.0",
     "@types/express": "^5.0.0",
     "@types/express-session": "^1.18.1",

package/routes/termdb.cluster.js

@@ -12,7 +12,6 @@ import { getData } from "#src/termdb.matrix.js";
 import { termType2label } from "#shared/terms.js";
 import { mayLog } from "#src/helpers.ts";
 import { formatElapsedTime } from "#shared/time.js";
-import { getResult as getResultGene } from "#src/gene.js";
 const api = {
   endpoint: "termdb/cluster",
   methods: {
@@ -220,26 +219,11 @@ 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";
 }
-async function validateHDF5File(filePath) {
-  try {
-    const jsonInput = JSON.stringify({
-      hdf5_file: filePath
-    });
-    const result = await run_rust("validateHDF5", jsonInput);
-    return JSON.parse(result);
-  } catch (error) {
-    console.error(`Error validating file: ${error}`);
-    return {
-      status: "error",
-      message: `Validation error`
-    };
-  }
-}
 async function queryGeneExpression(hdf5_file, geneNames) {
   const jsonInput = JSON.stringify({
     hdf5_file,
@@ -256,20 +240,21 @@ async function queryGeneExpression(hdf5_file, geneNames) {
     throw error;
   }
 }
-async function validateNative(q, ds, genome) {
+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);
+    await utils.file_is_readable(q.file);
+    const tmp = await run_rust("validateHDF5", JSON.stringify({ hdf5_file: q.file }));
+    const vr = JSON.parse(tmp);
     if (vr.status !== "success")
       throw vr.message;
     if (!vr.sampleNames?.length)
-      throw "HDF5 file has no samples";
+      throw "HDF5 file has no samples, please check file.";
     for (const sn of vr.sampleNames) {
       const si = ds.cohort.termdb.q.sampleName2id(sn);
       if (si == void 0)
-        throw "unknown sample from HDF5: " + sn;
+        throw `unknown sample ${sn} from HDF5 ${q.file}`;
       q.samples.push(si);
     }
     console.log(
@@ -277,97 +262,7 @@ async function validateNative(q, ds, genome) {
       vr.sampleNames.length
     );
   } catch (error) {
-    console.log(`${ds.label}: geneExpression HDF5 validation failed, falling back to tabix handling: ${error}`);
-    try {
-      q.samples = [];
-      await utils.validate_tabixfile(q.file);
-      q.nochr = await utils.tabix_is_nochr(q.file, null, genome);
-      const lines = await utils.get_header_tabix(q.file);
-      if (!lines[0])
-        throw "Header line missing from " + q.file;
-      const l = lines[0].split("	");
-      if (l.slice(0, 4).join("	") != "#chr	start	stop	gene") {
-        throw "Header line has wrong content for columns 1-4";
-      }
-      for (let i = 4; i < l.length; i++) {
-        const id = ds.cohort.termdb.q.sampleName2id(l[i]);
-        if (id == void 0) {
-          throw "queries.geneExpression: unknown sample from header: " + l[i];
-        }
-        q.samples.push(id);
-      }
-    } catch (e) {
-      throw `${ds.label} geneExpression file cannot be validated as tabix file`;
-    }
-    console.log(`${ds.label}: Tabix file successfully initialized. Samples: ${q.samples.length}`);
-    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 = {};
-      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;
-        }
-        if (!geneTerm.chr || !Number.isInteger(geneTerm.start) || !Number.isInteger(geneTerm.stop)) {
-          throw "Missing chr/start/stop";
-        }
-        const s2v = {};
-        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);
-        }
-      }
-      if (term2sample2value.size == 0) {
-        throw "No data available for the input " + param.terms?.map((g) => g.gene).join(", ");
-      }
-      return { term2sample2value, byTermId, bySampleId };
-    };
-    return;
+    throw `${ds.label}: Failed to validate geneExpression HDF5 file: ${error}`;
   }
   q.get = async (param) => {
     const limitSamples = await mayLimitSamples(param, q.samples, ds);
@@ -398,34 +293,30 @@ async function validateNative(q, ds, genome) {
       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)
+    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;
-        const geneResult = genesData[geneTerm.gene];
-        if (!geneResult) {
-          console.warn(`No data found for gene ${geneTerm.gene} in the response`);
+        if (limitSamples && !limitSamples.has(sampleId))
           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);
-        }
+        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(", ");