@sjcrh/proteinpaint-server 2.149.0 → 2.151.0

package/dataset/termdb.test.js

@@ -281,7 +281,8 @@ function termdb_test_default() {
       svfusion: {
         byrange: {
           file: "files/hg38/TermdbTest/TermdbTest_Fusion.gz"
-        }
+        },
+        dtLst: [2, 5]
       },
       cnv: {
         file: "files/hg38/TermdbTest/TermdbTest_CNV_gene.gz"
@@ -367,6 +368,20 @@ function termdb_test_default() {
       trackLst: {
         jsonFile: "files/hg38/TermdbTest/trackLst/facet.json",
         activeTracks: ["bw 1", "bed 1"]
+      },
+      chat: {},
+      alphaGenome: {
+        default: {
+          gene: "FLT3",
+          chromosome: "chr13",
+          position: 28034105,
+          reference: "A",
+          alternate: "AACTCCCATTTGAGATCATACC",
+          ontologyTerm: "EFO:0000572",
+          // lymphoblast
+          outputType: 4
+          //RNA SEQ
+        }
       }
     }
   };

package/genome/hg38.js

@@ -55,8 +55,8 @@ var hg38_default = {
     },
     {
       __isgene: true,
-      name: "GENCODE v47",
-      file: "anno/gencode.v47.hg38.gz",
+      name: "GENCODE v48",
+      file: "anno/gencode.v48.hg38.gz",
       translatecoding: true,
       categories: {
         coding: { color: "#004D99", label: "Coding gene" },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.149.0",
+  "version": "2.151.0",
   "type": "module",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "src/app.js",
@@ -61,11 +61,11 @@
   },
   "dependencies": {
     "@sjcrh/augen": "2.143.0",
-    "@sjcrh/proteinpaint-python": "2.146.0",
+    "@sjcrh/proteinpaint-python": "2.151.0",
     "@sjcrh/proteinpaint-r": "2.149.0",
-    "@sjcrh/proteinpaint-rust": "2.149.0",
-    "@sjcrh/proteinpaint-shared": "2.149.0",
-    "@sjcrh/proteinpaint-types": "2.149.0",
+    "@sjcrh/proteinpaint-rust": "2.150.0",
+    "@sjcrh/proteinpaint-shared": "2.150.0",
+    "@sjcrh/proteinpaint-types": "2.150.0",
     "@types/express": "^5.0.0",
     "@types/express-session": "^1.18.1",
     "better-sqlite3": "^9.4.1",

package/routes/alphaGenome.js

@@ -0,0 +1,41 @@
+import { alphaGenomePayload } from "#types/checkers";
+import { run_python } from "@sjcrh/proteinpaint-python";
+import serverconfig from "#src/serverconfig.js";
+const api = {
+  endpoint: "alphaGenome",
+  methods: {
+    get: {
+      ...alphaGenomePayload,
+      init
+    },
+    post: {
+      ...alphaGenomePayload,
+      init
+    }
+  }
+};
+function init() {
+  return async (req, res) => {
+    try {
+      const query = req.query;
+      const params = {
+        API_KEY: serverconfig.alphaGenome.API_KEY,
+        chromosome: query.chromosome,
+        position: query.position,
+        reference: query.reference,
+        alternate: query.alternate,
+        interval: query.interval
+      };
+      if (query.ontologyTerms) params["ontologyTerms"] = query.ontologyTerms;
+      if (query.outputType) params["outputType"] = Number(query.outputType);
+      const url = await run_python("alphaGenome.py", JSON.stringify(params));
+      res.send({ plotImage: url });
+    } catch (e) {
+      console.log(e);
+      res.status(404).send({ error: e });
+    }
+  };
+}
+export {
+  api
+};

package/routes/alphaGenomeTypes.js

@@ -0,0 +1,36 @@
+import { alphaGenomeTypesPayload } from "#types/checkers";
+import { run_python } from "@sjcrh/proteinpaint-python";
+import serverconfig from "#src/serverconfig.js";
+const api = {
+  endpoint: "AlphaGenomeTypes",
+  methods: {
+    get: {
+      ...alphaGenomeTypesPayload,
+      init
+    },
+    post: {
+      ...alphaGenomeTypesPayload,
+      init
+    }
+  }
+};
+function init() {
+  return async (req, res) => {
+    try {
+      const params = { API_KEY: serverconfig.alphaGenome.API_KEY };
+      const result = await run_python("AlphaGenomeTypes.py", JSON.stringify(params));
+      const { ontologyTerms, outputTypes, intervals } = JSON.parse(result);
+      res.send({
+        ontologyTerms: ontologyTerms.sort((a, b) => a.label.localeCompare(b.label)),
+        outputTypes,
+        intervals
+      });
+    } catch (e) {
+      console.log(e);
+      res.status(404).send({ error: e });
+    }
+  };
+}
+export {
+  api
+};

package/routes/grin2.js

@@ -5,7 +5,8 @@ import { run_python } from "@sjcrh/proteinpaint-python";
 import { mayLog } from "#src/helpers.ts";
 import { get_samples } from "#src/termdb.sql.js";
 import { read_file, file_is_readable } from "#src/utils.js";
-import { dtsnvindel, dtcnv, dtfusionrna } from "#shared/common.js";
+import { dtsnvindel, dtcnv, dtfusionrna, dtsv } from "#shared/common.js";
+import crypto from "crypto";
 const MAX_LESIONS_PER_TYPE = 5e4;
 const api = {
   endpoint: "grin2",
@@ -41,6 +42,20 @@ function init({ genomes }) {
     }
   };
 }
+function generateCacheFileName() {
+  const randomHex = crypto.randomBytes(16).toString("hex");
+  const cacheFileName = `grin2_results_${randomHex}.txt`;
+  return path.join(serverconfig.cachedir, "grin2", cacheFileName);
+}
+function getAvailableDataTypes(request) {
+  const availableOptions = [];
+  for (const key in request) {
+    if (key.endsWith("Options")) {
+      availableOptions.push(key);
+    }
+  }
+  return availableOptions;
+}
 async function runGrin2(g, ds, request) {
   const startTime = Date.now();
   const samples = await get_samples(
@@ -76,7 +91,10 @@ async function runGrin2(g, ds, request) {
     devicePixelRatio: request.devicePixelRatio,
     pngDotRadius: request.pngDotRadius,
     width: request.width,
-    height: request.height
+    height: request.height,
+    cacheFileName: generateCacheFileName(),
+    availableDataTypes: getAvailableDataTypes(request),
+    maxGenesToShow: request.maxGenesToShow
   };
   for (const c in g.majorchr) {
     if (ds.queries.singleSampleMutation.discoPlot?.skipChrM) {
@@ -112,7 +130,8 @@ async function runGrin2(g, ds, request) {
       grin2Time: grin2AnalysisTimeToPrint,
       totalTime
     },
-    processingSummary
+    processingSummary,
+    cacheFileName: resultData.cacheFileName
   };
   return response;
 }
@@ -121,6 +140,7 @@ function getLesionTracker(req) {
   if (req.snvindelOptions) currentTypes.push(dtsnvindel);
   if (req.cnvOptions) currentTypes.push(dtcnv);
   if (req.fusionOptions) currentTypes.push(dtfusionrna);
+  if (req.svOptions) currentTypes.push(dtsv);
   const track = /* @__PURE__ */ new Map();
   for (const t of currentTypes) track.set(t, { count: 0 });
   return track;
@@ -188,6 +208,9 @@ async function processSampleData(samples, ds, request, tracker) {
         case dtfusionrna:
           label = "fusion";
           break;
+        case dtsv:
+          label = "structural variant";
+          break;
         default:
           label = `type ${type}`;
           break;
@@ -237,8 +260,43 @@ async function processSampleMlst(sampleName, mlst, request, tracker) {
         }
         const les = filterAndConvertFusion(sampleName, m, request.fusionOptions);
         if (les && fusion) {
-          fusion.count++;
-          sampleLesions.push(les);
+          const lesionsToAdd = Array.isArray(les[0]) ? les.length : 1;
+          if (fusion.count + lesionsToAdd > MAX_LESIONS_PER_TYPE) {
+            break;
+          }
+          if (Array.isArray(les[0])) {
+            for (const lesion of les) {
+              sampleLesions.push(lesion);
+              fusion.count++;
+            }
+          } else {
+            sampleLesions.push(les);
+            fusion.count++;
+          }
+        }
+        break;
+      }
+      case dtsv: {
+        if (!request.svOptions) break;
+        const sv = tracker.get(dtsv);
+        if (sv && sv.count >= MAX_LESIONS_PER_TYPE) {
+          break;
+        }
+        const les = filterAndConvertSV(sampleName, m, request.svOptions);
+        if (les && sv) {
+          const lesionsToAdd = Array.isArray(les[0]) ? les.length : 1;
+          if (sv.count + lesionsToAdd > MAX_LESIONS_PER_TYPE) {
+            break;
+          }
+          if (Array.isArray(les[0])) {
+            for (const lesion of les) {
+              sampleLesions.push(lesion);
+              sv.count++;
+            }
+          } else {
+            sampleLesions.push(les);
+            sv.count++;
+          }
         }
         break;
       }
@@ -258,7 +316,20 @@ function filterAndConvertSnvIndel(sampleName, entry, options) {
   if (!Number.isInteger(entry.pos)) {
     return null;
   }
-  return [sampleName, entry.chr, entry.pos, entry.pos, "mutation"];
+  if (options.minAltAlleleCount !== void 0 && options.minAltAlleleCount > 0) {
+    if (!entry.altCount || entry.altCount < options.minAltAlleleCount) {
+      return null;
+    }
+  }
+  if (options.minTotalDepth !== void 0 && options.minTotalDepth > 0) {
+    const totalDepth = (entry.refCount || 0) + (entry.altCount || 0);
+    if (totalDepth < options.minTotalDepth) {
+      return null;
+    }
+  }
+  const start = entry.pos;
+  const end = entry.pos;
+  return [sampleName, entry.chr, start, end, "mutation"];
 }
 function filterAndConvertCnv(sampleName, entry, options) {
   if (!options || options.gainThreshold === void 0 || options.lossThreshold === void 0 || options.maxSegLength === void 0) {
@@ -277,10 +348,39 @@ function filterAndConvertCnv(sampleName, entry, options) {
   const isLoss = entry.value <= options.lossThreshold;
   if (!isGain && !isLoss) return null;
   const lesionType = isGain ? "gain" : "loss";
-  return [sampleName, entry.chr, entry.start, entry.stop, lesionType];
+  const start = entry.start;
+  const end = entry.stop;
+  return [sampleName, entry.chr, start, end, lesionType];
 }
 function filterAndConvertFusion(sampleName, entry, _options) {
-  return [sampleName, entry.chrA, entry.posA, entry.posB, "fusion"];
+  if (!entry.chrA || entry.posA === void 0) {
+    return null;
+  }
+  const startA = entry.posA;
+  const endA = entry.posA;
+  const lesionA = [sampleName, entry.chrA, startA, endA, "fusion"];
+  if (entry.chrB && entry.posB !== void 0) {
+    const startB = entry.posB;
+    const endB = entry.posB;
+    const lesionB = [sampleName, entry.chrB, startB, endB, "fusion"];
+    return [lesionA, lesionB];
+  }
+  return lesionA;
+}
+function filterAndConvertSV(sampleName, entry, _options) {
+  if (!entry.chrA || entry.posA === void 0) {
+    return null;
+  }
+  const startA = entry.posA;
+  const endA = entry.posA;
+  const lesionA = [sampleName, entry.chrA, startA, endA, "sv"];
+  if (entry.chrB && entry.posB !== void 0) {
+    const startB = entry.posB;
+    const endB = entry.posB;
+    const lesionB = [sampleName, entry.chrB, startB, endB, "sv"];
+    return [lesionA, lesionB];
+  }
+  return lesionA;
 }
 export {
   api

package/routes/termdb.DE.js

@@ -166,17 +166,24 @@ async function run_DE(param, ds, term_results, term_results2) {
   }
   const sample_size1 = group1names.length;
   const sample_size2 = group2names.length;
-  if (sample_size1 < 1) throw "sample size of group1 < 1";
-  if (sample_size2 < 1) throw "sample size of group2 < 1";
-  const commonnames = group1names.filter((element) => group2names.includes(element));
-  if (commonnames.length > 0) {
-    throw "Common elements found between both groups:" + commonnames.map((i) => i).join(",");
+  const alerts = validateGroups(sample_size1, sample_size2, group1names, group2names);
+  if (param.preAnalysis) {
+    const group1Name = param.samplelst.groups[0].name;
+    const group2Name = param.samplelst.groups[1].name;
+    return {
+      data: {
+        [group1Name]: sample_size1,
+        [group2Name]: sample_size2,
+        ...alerts.length ? { alert: alerts.join(" | ") } : {}
+      }
+    };
   }
-  const cases_string = group1names.map((i) => i).join(",");
-  const controls_string = group2names.map((i) => i).join(",");
+  if (alerts.length) throw alerts.join(" | ");
+  const cases_string = group2names.map((i) => i).join(",");
+  const controls_string = group1names.map((i) => i).join(",");
   const expression_input = {
-    case: controls_string,
-    control: cases_string,
+    case: cases_string,
+    control: controls_string,
     data_type: "do_DE",
     input_file: q.file,
     cachedir: serverconfig.cachedir,
@@ -236,6 +243,14 @@ async function run_DE(param, ds, term_results, term_results2) {
   param.method = "wilcoxon";
   return { data: result, sample_size1, sample_size2, method: param.method };
 }
+function validateGroups(sample_size1, sample_size2, group1names, group2names) {
+  const alerts = [];
+  if (sample_size1 < 1) alerts.push("sample size of group1 < 1");
+  if (sample_size2 < 1) alerts.push("sample size of group2 < 1");
+  const commonnames = group1names.filter((x) => group2names.includes(x));
+  if (commonnames.length) alerts.push(`Common elements found between both groups: ${commonnames.join(", ")}`);
+  return alerts;
+}
 async function readFileAndDelete(file, key, response) {
   const plot = await fs.promises.readFile(file);
   const plotBuffer = Buffer.from(plot).toString("base64");

package/routes/termdb.chat.js

@@ -23,6 +23,12 @@ function init({ genomes }) {
       if (!g) throw "invalid genome";
       const ds = g.datasets?.[q.dslabel];
       if (!ds) throw "invalid dslabel";
+      console.log("serverconfig:", serverconfig);
+      const serverconfig_ds_entries = serverconfig.genomes.find((genome) => genome.name == q.genome).datasets.find((dslabel) => dslabel.name == ds.label);
+      console.log("serverconfig_ds_entries:", serverconfig_ds_entries);
+      if (!serverconfig_ds_entries.aifiles) {
+        throw "aifiles are missing for chatbot to work";
+      }
       let apilink;
       let comp_model_name;
       let embedding_model_name;
@@ -41,11 +47,13 @@ function init({ genomes }) {
         // Just hardcoding variables here, these will later be defined in more appropriate places
         user_input: q.prompt,
         apilink,
-        dataset_db: serverconfig.tpmasterdir + "/" + ds.cohort.db.file,
+        tpmasterdir: serverconfig.tpmasterdir,
         comp_model_name,
         embedding_model_name,
-        llm_backend_name: serverconfig.llm_backend
+        llm_backend_name: serverconfig.llm_backend,
         // The type of backend (engine) used for running the embedding and completion model. Currently "SJ" and "Ollama" are supported
+        aifiles: serverconfig_ds_entries.aifiles,
+        binpath: serverconfig.binpath
       };
       const time1 = (/* @__PURE__ */ new Date()).valueOf();
       const ai_output_data = await run_rust("aichatbot", JSON.stringify(chatbot_input));

package/routes/termdb.config.js

@@ -73,6 +73,7 @@ function make(q, req, res, ds, genome) {
   if (tdb.excludedTermtypeByTarget) c.excludedTermtypeByTarget = tdb.excludedTermtypeByTarget;
   if (tdb.survival) c.survival = tdb.survival;
   if (tdb.regression) c.regression = tdb.regression;
+  if (tdb.uiLabels) c.uiLabels = tdb.uiLabels;
   if (ds.assayAvailability) c.assayAvailability = ds.assayAvailability;
   if (ds.cohort.correlationVolcano) c.correlationVolcano = ds.cohort.correlationVolcano;
   addRestrictAncestries(c, tdb);
@@ -135,7 +136,7 @@ function addNonDictionaryQueries(c, ds, genome) {
     if (q.snvindel.byisoform?.processTwsInOneQuery) q2.snvindel.byisoform = { processTwsInOneQuery: true };
   }
   if (q.svfusion) {
-    q2.svfusion = {};
+    q2.svfusion = { dtLst: q.svfusion.dtLst };
   }
   if (q.cnv) {
     q2.cnv = {};
@@ -180,6 +181,7 @@ function addNonDictionaryQueries(c, ds, genome) {
   if (q.chat) {
     q2.chat = {};
   }
+  if (q.alphaGenome) q2.alphaGenome = q.alphaGenome;
   if (q.NIdata && serverconfig.features.showBrainImaging) {
     q2.NIdata = {};
     for (const k in q.NIdata) {