@sjcrh/proteinpaint-server 2.57.0 → 2.58.0-0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.57.0",
+  "version": "2.58.0-0",
   "type": "module",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "src/app.js",
@@ -61,7 +61,7 @@
   },
   "dependencies": {
     "@sjcrh/augen": "2.46.0",
-    "@sjcrh/proteinpaint-rust": "2.55.0",
+    "@sjcrh/proteinpaint-rust": "2.58.0-0",
     "better-sqlite3": "^9.4.1",
     "body-parser": "^1.15.2",
     "canvas": "~2.11.2",

package/routes/genesetOverrepresentation.js

@@ -0,0 +1,56 @@
+import { run_rust } from "@sjcrh/proteinpaint-rust";
+const api = {
+  endpoint: "genesetOverrepresentation",
+  methods: {
+    all: {
+      init,
+      request: {
+        typeId: "genesetOverrepresentationRequest"
+      },
+      response: {
+        typeId: "genesetOverrepresentationResponse"
+        // will combine this with type checker
+        //valid: (t) => {}
+      }
+    }
+  }
+};
+function init({ genomes }) {
+  return async (req, res) => {
+    try {
+      const results = await run_genesetOverrepresentation_analysis(
+        req.query,
+        genomes
+      );
+      res.send(results);
+    } catch (e) {
+      res.send({ status: "error", error: e.message || e });
+    }
+  };
+}
+async function run_genesetOverrepresentation_analysis(q, genomes) {
+  if (!genomes[q.genome].termdbs)
+    throw "termdb database is not available for " + q.genome;
+  const gene_overrepresentation_input = {
+    sample_genes: q.sample_genes,
+    background_genes: q.background_genes,
+    db: genomes[q.genome].termdbs.msigdb.cohort.db.connection.name,
+    gene_set_group: q.geneSetGroup
+  };
+  const time1 = (/* @__PURE__ */ new Date()).valueOf();
+  const rust_output = await run_rust("genesetORA", JSON.stringify(gene_overrepresentation_input));
+  const time2 = (/* @__PURE__ */ new Date()).valueOf();
+  console.log("Time taken to run rust gene over representation pipeline:", time2 - time1, "ms");
+  let result;
+  for (const line of rust_output.split("\n")) {
+    if (line.startsWith("pathway_p_values:")) {
+      result = JSON.parse(line.replace("pathway_p_values:", ""));
+    } else {
+      console.log(line);
+    }
+  }
+  return result;
+}
+export {
+  api
+};

package/routes/hicdata.genome.js

@@ -0,0 +1,93 @@
+import { fileurl } from "#src/utils.js";
+import { spawn } from "child_process";
+import readline from "readline";
+import serverconfig from "#src/serverconfig.js";
+const api = {
+  endpoint: "hicgenome",
+  methods: {
+    get: {
+      init,
+      request: {
+        typeId: "HicdataRequest"
+      },
+      response: {
+        typeId: "HicdataResponse"
+      }
+    },
+    post: {
+      alternativeFor: "get",
+      init
+    }
+  }
+};
+function init() {
+  return async (req, res) => {
+    const data = [];
+    const erroutput = [];
+    const [e, file, isurl] = fileurl({ query: req.query });
+    if (e)
+      res.send({ error: "illegal file name" });
+    const matrixType = req.query.matrixType == "log(oe)" ? "oe" : req.query.matrixType ? req.query.matrixType : "observed";
+    const promises = req.query.chrlst.map((lead, i) => {
+      return req.query.chrlst.slice(0, i + 1).map((follow, j) => {
+        if (j <= i) {
+          return new Promise((resolve, reject) => {
+            const pos1 = req.query.nochr ? lead.replace("chr", "") : lead;
+            const pos2 = req.query.nochr ? follow.replace("chr", "") : follow;
+            const par = [
+              matrixType,
+              req.query.nmeth || "NONE",
+              file,
+              pos1,
+              pos2,
+              req.query.isfrag ? "FRAG" : "BP",
+              req.query.resolution
+            ];
+            const ps = spawn(serverconfig.hicstraw, par);
+            const rl = readline.createInterface({ input: ps.stdout });
+            const items = [];
+            let linenot3fields = 0;
+            let fieldnotnumerical = 0;
+            rl.on("line", (line) => {
+              const l = line.split("	");
+              if (l.length != 3) {
+                linenot3fields++;
+                return;
+              }
+              const n1 = Number.parseInt(l[0]);
+              const n2 = Number.parseInt(l[1]);
+              const v = req.query.matrixType == "log(oe)" ? Math.log(Number.parseFloat(l[2])) : Number.parseFloat(l[2]);
+              if (Number.isNaN(n1) || Number.isNaN(n2) || Number.isNaN(v)) {
+                fieldnotnumerical++;
+                return;
+              }
+              if (req.query.mincutoff != void 0 && v <= req.query.mincutoff) {
+                return;
+              }
+              items.push([n1, n2, v]);
+            });
+            data.push({ items, lead, follow });
+            ps.stderr.on("data", (i2) => erroutput.push(`${lead} - ${follow}: `, i2));
+            ps.on("close", () => {
+              if (erroutput.length)
+                reject({ error: erroutput.join("") });
+              if (linenot3fields)
+                reject({ error: `${linenot3fields} lines have other than 3 fields` });
+              if (fieldnotnumerical)
+                reject(`${fieldnotnumerical} lines have non-numerical values in any of the 3 fields`);
+              resolve();
+            });
+          });
+        }
+      });
+    }).flat();
+    Promise.allSettled(promises).then(() => res.send({ data, error: erroutput.join("") })).catch((e2) => {
+      res.send({ error: e2?.message || e2 });
+      if (e2 instanceof Error && e2.stack)
+        console.log(e2);
+    });
+  };
+}
+export {
+  api
+};

package/routes/termdb.categories.js

@@ -87,10 +87,23 @@ function init({ genomes }) {
 async function trigger_getcategories(q, res, tdb, ds, genome) {
   if (!q.tid)
     throw ".tid missing";
-  const term = q.type == "geneVariant" ? { name: q.tid, type: "geneVariant", isleaf: true } : tdb.q.termjsonByOneid(q.tid);
+  let term;
+  if (q.type == "geneVariant") {
+    term = { id: q.name, name: q.name, type: "geneVariant", isleaf: true };
+    if (q.gene) {
+      term.gene = q.gene;
+    } else {
+      term.chr = q.chr;
+      term.start = q.start;
+      term.stop = q.stop;
+    }
+  } else {
+    term = tdb.q.termjsonByOneid(q.tid);
+  }
+  const tw = q.type == "geneVariant" ? { $id: q.gene || q.name || q.tid, term, q: { isAtomic: true } } : { $id: q.tid, id: q.tid, term, q: q.term1_q || getDefaultQ(term, q) };
   const arg = {
     filter: q.filter,
-    terms: q.type == "geneVariant" ? [{ term, q: { isAtomic: true } }] : [{ id: q.tid, term, q: q.term1_q || getDefaultQ(term, q) }],
+    terms: [tw],
     currentGeneNames: q.currentGeneNames,
     // optional, from mds3 mayAddGetCategoryArgs()
     rglst: q.rglst
@@ -112,7 +125,8 @@ async function trigger_getcategories(q, res, tdb, ds, genome) {
     }
     const sampleCountedFor = /* @__PURE__ */ new Set();
     for (const [sampleId, sampleData] of Object.entries(samples)) {
-      const values = sampleData[q.tid].values;
+      const key = tw.$id;
+      const values = sampleData[key].values;
       sampleCountedFor.clear();
       for (const value of values) {
         if (!dtClassMap.has(value.dt)) {

package/routes/termdb.cluster.js

@@ -136,6 +136,7 @@ async function validate_query_geneExpression(ds, genome) {
   const q = ds.queries.geneExpression;
   if (!q)
     return;
+  q.gene2density = {};
   if (q.src == "gdcapi") {
     gdc_validate_query_geneExpression(ds, genome);
     return;

package/routes/termdb.config.js

@@ -206,6 +206,8 @@ function getAllowedTermTypes(ds) {
   if (ds?.queries?.defaultBlock2GeneMode) {
     typeSet.add("geneVariant");
   }
+  if (ds?.queries?.geneExpression)
+    typeSet.add("geneExpression");
   return [...typeSet];
 }
 export {