@sjcrh/proteinpaint-server 2.146.0 → 2.146.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.146.0",
+  "version": "2.146.2",
   "type": "module",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "src/app.js",
@@ -64,8 +64,8 @@
     "@sjcrh/proteinpaint-python": "2.146.0",
     "@sjcrh/proteinpaint-r": "2.145.1-0",
     "@sjcrh/proteinpaint-rust": "2.146.0",
-    "@sjcrh/proteinpaint-shared": "2.145.2",
-    "@sjcrh/proteinpaint-types": "2.146.0",
+    "@sjcrh/proteinpaint-shared": "2.146.2",
+    "@sjcrh/proteinpaint-types": "2.146.2",
     "@types/express": "^5.0.0",
     "@types/express-session": "^1.18.1",
     "better-sqlite3": "^9.4.1",

package/routes/correlationVolcano.js

@@ -2,7 +2,7 @@ import { CorrelationVolcanoPayload } from "#types/checkers";
 import { getData } from "../src/termdb.matrix.js";
 import { run_R } from "@sjcrh/proteinpaint-r";
 import { mayLog } from "#src/helpers.ts";
-import { getStdDev } from "#shared/descriptive.stats.js";
+import { getStdDev } from "./termdb.descrstats.ts";
 const minArrayLength = 3;
 const api = {
   endpoint: "termdb/correlationVolcano",

package/routes/termdb.boxplot.js

@@ -2,7 +2,7 @@ import { boxplotPayload } from "#types/checkers";
 import { getData } from "../src/termdb.matrix.js";
 import { boxplot_getvalue } from "../src/utils.js";
 import { sortPlot2Values } from "./termdb.violin.ts";
-import { summaryStats, getDescriptiveStats, summaryStatsFromStats } from "#shared/descriptive.stats.js";
+import { getDescrStats } from "./termdb.descrstats.ts";
 const api = {
   endpoint: "termdb/boxplot",
   methods: {
@@ -30,8 +30,8 @@ function init({ genomes }) {
       const data = await getData({ filter: q.filter, filter0: q.filter0, terms, __protected__: q.__protected__ }, ds);
       if (data.error) throw data.error;
       const samples = Object.values(data.samples);
-      const values = samples.map((s) => s?.[q.tw.$id]?.value).filter((v) => typeof v === "number");
-      const statsAllSummary = summaryStats(values).values;
+      const values = samples.map((s) => s?.[q.tw.$id]?.value).filter((v) => typeof v === "number" && !q.tw.term.values?.[v]?.uncomputable);
+      const descrStats = getDescrStats(values, q.removeOutliers);
       const sampleType = `All ${data.sampleType?.plural_name || "samples"}`;
       const overlayTerm = q.overlayTw;
       const divideTerm = q.divideTw;
@@ -55,12 +55,10 @@ function init({ genomes }) {
             const value = { value: v };
             return value;
           });
-          const stats = getDescriptiveStats(sortedValues);
           if (q.removeOutliers) {
-            outlierMin = Math.min(outlierMin, stats.outlierMin);
-            outlierMax = Math.max(outlierMax, stats.outlierMax);
+            outlierMin = Math.min(outlierMin, descrStats.outlierMin.value);
+            outlierMax = Math.max(outlierMax, descrStats.outlierMax.value);
           }
-          const descrStats = summaryStatsFromStats(stats, true).values;
           const boxplot = boxplot_getvalue(vs, q.removeOutliers);
           if (!boxplot) throw "boxplot_getvalue failed [termdb.boxplot init()]";
           const _plot = {
@@ -100,7 +98,7 @@ function init({ genomes }) {
         absMax: q.removeOutliers ? outlierMax : absMax,
         charts,
         uncomputableValues: setUncomputableValues(uncomputableValues),
-        descrStats: statsAllSummary
+        descrStats
       };
       res.send(returnData);
     } catch (e) {

package/routes/termdb.cohort.summary.js

@@ -20,7 +20,7 @@ function init({ genomes }) {
       if (!genome) throw "invalid genome";
       const [ds] = get_ds_tdb(genome, q);
       let count;
-      if (ds.cohort.termdb.getAdditionalFilter) {
+      if (q.filter?.lst?.length) {
         const samples = await get_samples(q, ds);
         count = samples.length;
       } else count = ds.cohort.termdb.q?.getCohortSampleCount?.(q.cohort) || 1;

package/routes/termdb.descrstats.js

@@ -1,6 +1,7 @@
 import { descrStatsPayload } from "#types/checkers";
-import { summaryStats } from "#shared/descriptive.stats.js";
 import { getData } from "#src/termdb.matrix.js";
+import computePercentile from "#shared/compute.percentile.js";
+import { roundValueAuto } from "#shared/roundValue.js";
 const api = {
   endpoint: "termdb/descrstats",
   methods: {
@@ -17,7 +18,6 @@ const api = {
 function init({ genomes }) {
   return async (req, res) => {
     const q = req.query;
-    let result;
     try {
       const genome = genomes[req.query.genome];
       if (!genome) throw "invalid genome name";
@@ -26,41 +26,98 @@ function init({ genomes }) {
       const tdb = ds.cohort.termdb;
       if (!tdb) throw "invalid termdb object";
       if (!q.tw.$id) q.tw.$id = "_";
-      const data = await getData(
-        { filter: q.filter, filter0: q.filter0, terms: [q.tw], __protected__: q.__protected__ },
-        ds
-      );
-      if (data.error) throw data.error;
-      const values = [];
-      for (const key in data.samples) {
-        const sample = data.samples[key];
-        const v = sample[q.tw.$id];
-        if (!v && v !== 0) {
-          continue;
-        }
-        const value = v.value;
-        if (q.tw.q.hiddenValues?.[value]) {
-          continue;
-        }
-        if (q.logScale) {
-          if (value === 0) {
-            continue;
-          }
-        }
-        values.push(Number(value));
-      }
-      if (values.length) {
-        result = summaryStats(values);
-      } else {
-        result = {};
-      }
+      const result = await trigger_getDescrStats(q, ds);
+      res.send(result);
     } catch (e) {
       if (e instanceof Error && e.stack) console.log(e);
-      result = { error: e?.message || e };
+      const result = { error: e?.message || e };
+      res.send(result);
+    }
+  };
+}
+async function trigger_getDescrStats(q, ds) {
+  const data = await getData(
+    { filter: q.filter, filter0: q.filter0, terms: [q.tw], __protected__: q.__protected__ },
+    ds
+  );
+  if (data.error) throw data.error;
+  const values = [];
+  for (const key in data.samples) {
+    const sample = data.samples[key];
+    const v = sample[q.tw.$id];
+    if (!v && v !== 0) {
+      continue;
+    }
+    const value = v.value;
+    if (q.tw.term.values?.[value]?.uncomputable) {
+      continue;
+    }
+    if (q.logScale) {
+      if (value === 0) {
+        continue;
+      }
     }
-    res.send(result);
+    values.push(Number(value));
+  }
+  const stats = getDescrStats(values);
+  return stats;
+}
+function getDescrStats(values, showOutlierRange = false) {
+  if (!values.length) {
+    return {};
+  }
+  if (values.some((v) => !Number.isFinite(v))) throw "non-numeric values found";
+  const sorted_arr = values.sort((a, b) => a - b);
+  const n = sorted_arr.length;
+  const median = computePercentile(sorted_arr, 50, true);
+  const mean = getMean(sorted_arr);
+  const variance = getVariance(sorted_arr);
+  const stdDev = Math.sqrt(variance);
+  const p25 = computePercentile(sorted_arr, 25, true);
+  const p75 = computePercentile(sorted_arr, 75, true);
+  const IQR = p75 - p25;
+  const min = sorted_arr[0];
+  const max = sorted_arr[sorted_arr.length - 1];
+  const outlierMin = p25 - 1.5 * IQR;
+  const outlierMax = p75 + 1.5 * IQR;
+  const stats = {
+    total: { key: "total", label: "Total", value: n },
+    min: { key: "min", label: "Minimum", value: min },
+    p25: { key: "p25", label: "1st quartile", value: p25 },
+    median: { key: "median", label: "Median", value: median },
+    p75: { key: "p75", label: "3rd quartile", value: p75 },
+    max: { key: "max", label: "Maximum", value: max },
+    mean: { key: "mean", label: "Mean", value: mean },
+    stdDev: { key: "stdDev", label: "Standard deviation", value: stdDev }
+    //variance: { label: 'Variance', value: variance }, // not necessary to report, as it is just stdDev^2
+    //iqr: { label: 'Inter-quartile range', value: IQR } // not necessary to report, as it is just p75-p25
   };
+  if (showOutlierRange) {
+    stats.outlierMin = { key: "outlierMin", label: "Outlier minimum", value: outlierMin };
+    stats.outlierMax = { key: "outlierMax", label: "Outlier maximum", value: outlierMax };
+  }
+  for (const v of Object.values(stats)) {
+    const rounded = roundValueAuto(v.value);
+    v.value = rounded;
+  }
+  return stats;
+}
+function getMean(data) {
+  return data.reduce((sum, value) => sum + value, 0) / data.length;
+}
+function getVariance(data) {
+  const meanValue = getMean(data);
+  const squaredDifferences = data.map((value) => Math.pow(value - meanValue, 2));
+  return squaredDifferences.reduce((sum, value) => sum + value, 0) / (data.length - 1);
+}
+function getStdDev(data) {
+  const variance = getVariance(data);
+  return Math.sqrt(variance);
 }
 export {
-  api
+  api,
+  getDescrStats,
+  getMean,
+  getStdDev,
+  getVariance
 };

package/routes/termdb.violin.js

@@ -4,7 +4,7 @@ import { run_rust } from "@sjcrh/proteinpaint-rust";
 import { getData } from "../src/termdb.matrix.js";
 import { createCanvas } from "canvas";
 import { getOrderedLabels } from "../src/termdb.barchart.js";
-import { summaryStats } from "#shared/descriptive.stats.js";
+import { getDescrStats } from "./termdb.descrstats.ts";
 import { isNumericTerm } from "#shared/terms.js";
 import { numericBins, parseValues } from "./termdb.boxplot.ts";
 import { run_R } from "@sjcrh/proteinpaint-r";
@@ -57,9 +57,9 @@ async function trigger_getViolinPlotData(q, ds) {
     ds
   );
   const samples = Object.values(data.samples);
-  let values = samples.map((s) => s?.[q.tw.$id]?.value).filter((v) => typeof v === "number");
+  let values = samples.map((s) => s?.[q.tw.$id]?.value).filter((v) => typeof v === "number" && !q.tw.term.values?.[v]?.uncomputable);
   if (q.unit == "log") values = values.filter((v) => v > 0);
-  const descrStats = summaryStats(values).values;
+  const descrStats = getDescrStats(values);
   const sampleType = `All ${data.sampleType?.plural_name || "samples"}`;
   if (data.error) throw data.error;
   if (q.overlayTw && data.refs.byTermId[q.overlayTw.$id]) {
@@ -225,7 +225,7 @@ async function createCanvasImg(q, result, ds) {
           ctx.stroke();
         });
       plot.src = canvas.toDataURL();
-      plot.summaryStats = summaryStats(plot.values).values;
+      plot.summaryStats = getDescrStats(plot.values);
     }
   }
 }