@sjcrh/proteinpaint-server 2.186.0 → 2.188.0

package/routes/termdb.proteome.js

@@ -42,12 +42,12 @@ function init({ genomes }) {
               term: {
                 name: term.name,
                 type: "proteomeAbundance",
-                proteomeDetails: details
+                dataTypeDetails: details
               }
             };
             const cohortData = await ds.queries.proteome.get({
               terms: [tw],
-              proteomeDetails: details,
+              dataTypeDetails: details,
               filter: q.filter,
               filter0: q.filter0,
               for: "proteinView",
@@ -64,6 +64,7 @@ function init({ genomes }) {
               entry.testedN = stats.testedN;
               entry.controlN = stats.controlN;
               if (assay.mclassOverride) entry.mclassOverride = assay.mclassOverride;
+              if (organism.genomeName) entry.genomeName = organism.genomeName;
               cohorts.push(entry);
             }
           }
@@ -248,7 +249,7 @@ async function validate_query_proteome(ds) {
     const proteins = arg?.proteins;
     if (!Array.isArray(proteins) || proteins.length == 0) throw "queries.proteome.find arg.proteins[] missing";
     const matches = /* @__PURE__ */ new Set();
-    const details = arg?.proteomeDetails || {};
+    const details = arg?.dataTypeDetails || {};
     const organism = details.organism;
     const assay = details.assay;
     const cohort = details.cohort;
@@ -256,7 +257,7 @@ async function validate_query_proteome(ds) {
     const filters = [];
     if (Object.keys(details).length) {
       if (!organism || !assay || !cohort)
-        throw "queries.proteome.find arg.proteomeDetails.{organism,assay,cohort} missing";
+        throw "queries.proteome.find arg.dataTypeDetails.{organism,assay,cohort} missing";
       const organismConfig = q.organisms?.[organism];
       if (!organismConfig) throw `queries.proteome.find invalid organism: ${organism}`;
       const assayConfig = organismConfig.assays?.[assay];
@@ -299,8 +300,8 @@ async function validate_query_proteome(ds) {
   };
   q.get = async (param) => {
     if (!param?.terms?.length) throw "queries.proteome.get param.terms[] missing";
-    if (!param.proteomeDetails?.assay || !param.proteomeDetails?.cohort || !param.proteomeDetails?.organism)
-      throw "queries.proteome.get param.proteomeDetails.{assay,cohort,organism} missing";
+    if (!param.dataTypeDetails?.assay || !param.dataTypeDetails?.cohort || !param.dataTypeDetails?.organism)
+      throw "queries.proteome.get param.dataTypeDetails.{assay,cohort,organism} missing";
     return await getProteomeValuesFromCohort(ds, param, q);
   };
 }
@@ -328,6 +329,12 @@ function buildFilterClause(filters) {
   }
   return { conditions, params };
 }
+function countDistinctSamples(db, filters) {
+  if (!filters?.length) throw "countDistinctSamples: filters must not be empty";
+  const { conditions, params } = buildFilterClause(filters);
+  const row = db.prepare(`SELECT COUNT(DISTINCT sample) as cnt FROM proteome_abundance WHERE ${conditions.join(" AND ")}`).get(...params);
+  return row?.cnt || 0;
+}
 function queryDbRows(db, matchColumn, matchValue, filters) {
   const { conditions, params } = buildFilterClause(filters);
   const allConditions = [`${matchColumn} = ? COLLATE NOCASE`, ...conditions];
@@ -338,7 +345,7 @@ function queryDbRows(db, matchColumn, matchValue, filters) {
 }
 async function getProteomeValuesFromCohort(ds, param, q) {
   const db = ds.queries.proteome.db;
-  const { assay, cohort, organism } = param.proteomeDetails;
+  const { assay, cohort, organism } = param.dataTypeDetails;
   const organismConfig = q.organisms?.[organism];
   if (!organismConfig) throw `queries.proteome invalid organism: ${organism}`;
   const organismColumnIdx = organismConfig.columnIdx;
@@ -452,5 +459,6 @@ async function getProteomeValuesFromCohort(ds, param, q) {
 }
 export {
   api,
+  countDistinctSamples,
   validate_query_proteome
 };

package/routes/termdb.sampleScatter.js

@@ -53,6 +53,10 @@ function init({ genomes }) {
         cohortSamples = tmp[0];
       } else {
         if (!q.plotName) throw new Error("Neither plot name or coordinates where provided");
+        if (typeof ds.cohort?.scatterplots?.get == "function") {
+          const allowed = ds.cohort.scatterplots.get(q.__protected__?.clientAuthResult);
+          if (!allowed?.find((i) => i.name == q.plotName)) throw new Error("No permission to display plot");
+        }
         if (!Array.isArray(ds.cohort?.scatterplots?.plots)) throw new Error("not supported");
         const plot = ds.cohort.scatterplots.plots.find((p) => p.name == q.plotName);
         if (!plot) throw new Error(`plot not found with plotName ${q.plotName}`);

package/routes/termdb.singlecellSamples.js

@@ -89,6 +89,17 @@ async function validateSamples(q, ds) {
   const S = q.samples, D = q.data;
   const samples = /* @__PURE__ */ new Map();
   for (const plot of D.plots) {
+    if (plot.isMetaResult) {
+      const sampleName = plot?.sampleId || plot.name.replace(/\s/g, "_");
+      const tsvfile = path.join(serverconfig.tpmasterdir, plot.folder, sampleName + (plot.fileSuffix || ""));
+      try {
+        await file_is_readable(tsvfile);
+        samples.set(sampleName, { sample: sampleName, isMetaResult: true });
+      } catch (e) {
+        throw new Error(`meta result data file missing or unreadable: ${sampleName} (${tsvfile}): ${e.message || e}`);
+      }
+      continue;
+    }
     for (const fn of await fs.promises.readdir(path.join(serverconfig.tpmasterdir, plot.folder))) {
       let sampleName = fn;
       if (plot.fileSuffix) {

package/routes/termdb.topVariablyExpressedGenes.js

@@ -1,11 +1,7 @@
 import { termdbTopVariablyExpressedGenesPayload } from "#types/checkers";
-import { run_rust } from "@sjcrh/proteinpaint-rust";
-import serverconfig from "#src/serverconfig.js";
 import { mayLimitSamples } from "#src/mds3.filter.js";
-import { makeFilter } from "#src/mds3.gdc.js";
-import { cachedFetch } from "#src/utils.js";
-import { joinUrl } from "#shared/joinUrl.js";
-import { formatElapsedTime } from "#shared/time.js";
+import { run_python } from "@sjcrh/proteinpaint-python";
+import { mayLog } from "#src/helpers.ts";
 const api = {
   endpoint: "termdb/topVariablyExpressedGenes",
   methods: {
@@ -29,27 +25,23 @@ function init({ genomes }) {
       const ds = genome.datasets?.[q.dslabel];
       if (!ds) throw "invalid dslabel";
       if (!ds.queries?.topVariablyExpressedGenes) throw "not supported on dataset";
+      q.ds = ds;
       const t = Date.now();
       result = {
         genes: await ds.queries.topVariablyExpressedGenes.getGenes(q)
       };
-      console.log("compute top variably expressed genes:", formatElapsedTime(Date.now() - t));
+      mayLog("time for top variably expressed genes", Date.now() - t);
     } catch (e) {
       result = { status: e.status || 400, error: e.message || e };
     }
     res.send(result);
   };
 }
-function validate_query_TopVariablyExpressedGenes(ds, genome) {
+function validate_query_TopVariablyExpressedGenes(ds) {
   const q = ds.queries.topVariablyExpressedGenes;
   if (!q) return;
-  if (q.src == "gdcapi") {
-    gdcValidateQuery(ds, genome);
-  } else if (q.src == "native") {
-    nativeValidateQuery(ds);
-  } else {
-    throw "unknown topVariablyExpressedGenes.src";
-  }
+  if (typeof q.getGenes == "function") return;
+  nativeValidateQuery(ds);
 }
 function nativeValidateQuery(ds) {
   const gE = ds.queries.geneExpression;
@@ -83,21 +75,7 @@ function addTopVEarg(q) {
       id: "filter_extreme_values",
       label: "Filter Extreme Values",
       type: "boolean",
-      value: true,
-      options: [
-        {
-          id: "min_count",
-          label: "Min count",
-          type: "number",
-          value: 10
-        },
-        {
-          id: "min_total_count",
-          label: "Min total count",
-          type: "number",
-          value: 15
-        }
-      ]
+      value: true
     },
     {
       id: "rank_type",
@@ -134,124 +112,20 @@ function addTopVEarg(q) {
   q.arguments = arglst;
 }
 async function computeGenes4nativeDs(q, gE, samples) {
+  if (!["number", "boolean"].includes(typeof q.filter_extreme_values) || q.filter_extreme_values === void 0) {
+    q.filter_extreme_values = false;
+  }
   const input_json = {
     input_file: gE.file,
     samples: samples.join(","),
-    filter_extreme_values: q.filter_extreme_values,
-    num_genes: q.maxGenes,
-    rank_type: q.rank_type?.type
+    filter_extreme_values: typeof q.filter_extreme_values === "number" ? Boolean(q.filter_extreme_values) : q.filter_extreme_values,
+    max_genes: q.maxGenes,
+    rank_type: q.rank_type?.type ?? "var"
   };
-  if (q.filter_extreme_values == 1) {
-    input_json["min_count"] = q.min_count;
-    input_json["min_total_count"] = q.min_total_count;
-  }
-  if (gE.newformat) {
-    input_json["newformat"] = true;
-  }
-  const rust_output = await run_rust("topGeneByExpressionVariance", JSON.stringify(input_json));
-  const rust_output_list = rust_output.split("\n");
-  let output_json;
-  for (const item of rust_output_list) {
-    if (item.includes("output_json:")) {
-      output_json = JSON.parse(item.replace("output_json:", ""));
-    } else {
-      console.log(item);
-    }
-  }
-  const varGenes = output_json.map((i) => i.gene_symbol);
+  const python_output = await run_python("topVEgene.py", JSON.stringify(input_json));
+  const varGenes = typeof python_output === "string" ? JSON.parse(python_output) : [];
   return varGenes;
 }
-function gdcValidateQuery(ds, genome) {
-  ds.queries.topVariablyExpressedGenes.getGenes = async (q) => {
-    if (serverconfig.features.gdcGenes) {
-      console.error(
-        "!!GDC!! using serverconfig.features.gdcGenes[] but not live api query. only use this on DEV and never on PROD!"
-      );
-      return serverconfig.features.gdcGenes;
-    }
-    if (ds.label === "GDC" && !ds.__gdc?.doneCaching) {
-      throw "The server has not finished caching the case IDs: try again in about 2 minutes.";
-    }
-    const { host, headers } = ds.getHostHeaders(q);
-    try {
-      const response = await cachedFetch(
-        joinUrl(host.rest, "/gene_expression/gene_selection"),
-        {
-          method: "POST",
-          headers,
-          body: getGeneSelectionArg(q)
-        },
-        {
-          // noCache: true, // !!! for testing only !!!
-          getErrMessage: (response2) => {
-            const body = response2?.body || response2;
-            return Array.isArray(body?.gene_selection) ? "" : body?.message || body?.error || JSON.stringify(body);
-          }
-        }
-      );
-      const re = response.body;
-      const genes = [];
-      if (!Array.isArray(re.gene_selection)) {
-        throw "re.gene_selection[] is not array: " + JSON.stringify(re);
-      }
-      for (const i of re.gene_selection) {
-        if (i.gene_id && typeof i.gene_id == "string") {
-          const t = genome.genedb.getNameByAlias.get(i.gene_id);
-          if (t) genes.push(t.name);
-        } else if (i.symbol && typeof i.symbol == "string") {
-          genes.push(i.symbol);
-        } else {
-          throw "one of re.gene_selection[] is missing both gene_id and symbol";
-        }
-      }
-      return genes;
-    } catch (e) {
-      console.error(e.stack || e);
-      throw e;
-    }
-  };
-  function getGeneSelectionArg(q) {
-    const arg = {
-      // add any to avoid tsc err
-      case_filters: makeFilter(q),
-      selection_size: q.maxGenes,
-      min_median_log2_uqfpkm: q.min_median_log2_uqfpkm
-    };
-    if (q.geneSet) {
-      if (q.geneSet.type == "all") {
-        arg.gene_type = "protein_coding";
-      } else if (q.geneSet.type == "custom" || q.geneSet.type == "msigdb") {
-        if (!Array.isArray(q.geneSet.geneList)) throw "q.geneSet.geneList is not array";
-        arg.gene_ids = map2ensg(q.geneSet.geneList, genome);
-        if (arg.gene_ids.length == 0) throw "no valid genes from custom gene set";
-      } else {
-        throw "unknown q.geneSet.type";
-      }
-    } else {
-      arg.gene_type = "protein_coding";
-    }
-    return arg;
-  }
-}
-function map2ensg(lst, genome) {
-  const ensg = [];
-  for (const name of lst) {
-    if (name.startsWith("ENSG") && name.length == 15) {
-      ensg.push(name);
-      continue;
-    }
-    const tmp = genome.genedb.getAliasByName.all(name);
-    if (Array.isArray(tmp)) {
-      for (const a of tmp) {
-        if (a.alias.startsWith("ENSG")) {
-          ensg.push(a.alias);
-          break;
-        }
-      }
-    }
-  }
-  return ensg;
-}
 export {
   api,
   validate_query_TopVariablyExpressedGenes