feat: cdf loader for gcms added (#290)

* feat: cdf loader for gcms added

* feat: add cdfLoader in the app

* test: add direct link to the test files HPLC and GC/MS

* chore: specify the return value of loaders

* chore: fixing missing importt

* chore: suggestions implemented

* fix: style issues fixed

* fix: return full experiment as measurement

* chore: fix the code

Co-authored-by: Luc Patiny <luc@patiny.com>

src/app/context/load.ts

@@ -4,6 +4,7 @@ import { loadMeasurements } from '../data/append';
 import { getIRAutoPeakPickingEnhancer } from '../data/enhancers/irAutoPeakPickingEnhancer';
 import { irMeasurementEnhancer } from '../data/enhancers/irMeasurementEnhancer';
 import { biologicLoader } from '../data/loaders/biologicLoader';
+import { cdfLoader } from '../data/loaders/cdfLoader';
 import { jcampLoader } from '../data/loaders/jcampLoader';
 import { cary500Loader } from '../data/loaders/proprietary/agilent/cary500Loader';
 import { spcLoader } from '../data/loaders/spcLoader';
@@ -12,7 +13,14 @@ import { wdfLoader } from '../data/loaders/wdfLoader';
 import type { AppDispatch } from './appState';
 
 const options = {
-  loaders: [jcampLoader, spcLoader, wdfLoader, biologicLoader, cary500Loader],
+  loaders: [
+    jcampLoader,
+    spcLoader,
+    wdfLoader,
+    biologicLoader,
+    cary500Loader,
+    cdfLoader,
+  ],
   enhancers: {
     ir: [
       irMeasurementEnhancer,

src/app/data/__tests__/getGCMSMeasurement.test.ts

@@ -0,0 +1,8 @@
+import { expect, test } from 'vitest';
+
+import { getGCMSMeasurement } from './getGCMSMeasurement';
+
+test('getGCMSMeasurement', async () => {
+  const result = await getGCMSMeasurement();
+  expect(result).toHaveLength(1);
+});

src/app/data/__tests__/getGCMSMeasurement.ts

@@ -0,0 +1,20 @@
+import { join } from 'path';
+
+import { fileCollectionFromPath } from 'filelist-utils';
+
+import { getEmptyDataState } from '../DataState';
+import { append } from '../append';
+import { cdfLoader } from '../loaders/cdfLoader';
+
+export async function getGCMSMeasurement() {
+  const dataState = getEmptyDataState();
+
+  const filteredFileCollection = (
+    await fileCollectionFromPath(join(__dirname, 'data/cdf/'))
+  ).filter((file) => file.name === 'agilent-gcms.cdf');
+
+  const gcmsEntry = (
+    await append(filteredFileCollection, dataState, { loaders: [cdfLoader] })
+  ).dataState.measurements.gclcms.entries;
+  return gcmsEntry;
+}

src/app/data/loaders/biologicLoader.ts

@@ -2,12 +2,12 @@ import { v4 } from '@lukeed/uuid';
 import { convert } from 'biologic-converter';
 import type { FileCollection } from 'filelist-utils';
 
-import { getEmptyMeasurements, Loader } from '../DataState';
+import { getEmptyMeasurements, Loader, Measurements } from '../DataState';
 import type { MeasurementBase } from '../MeasurementBase';
 
 export const biologicLoader: Loader = async function biologicLoader(
   fileCollection: FileCollection,
-) {
+): Promise<Measurements> {
   let measurements = getEmptyMeasurements();
   const results = await convert(fileCollection);
   for (let { dir, mpr, mps, mpt } of results) {

src/app/data/loaders/cdfLoader.ts

@@ -0,0 +1,96 @@
+import { v4 } from '@lukeed/uuid';
+import type { FileCollection } from 'filelist-utils';
+import { NetCDFReader } from 'netcdfjs';
+
+import {
+  MeasurementKind,
+  Loader,
+  Measurements,
+  getEmptyMeasurements,
+} from '../DataState';
+
+export const cdfLoader: Loader = async function cdfLoader(
+  fileCollection: FileCollection,
+): Promise<Measurements> {
+  const newMeasurements: Measurements = getEmptyMeasurements();
+  for (const file of fileCollection) {
+    if (file.name.match(/(?:\.cdf)$/i)) {
+      const reader = new NetCDFReader(await file.arrayBuffer(), { meta: true });
+      if (
+        !reader.dataVariableExists('mass_values') ||
+        !reader.dataVariableExists('time_values')
+      ) {
+        return newMeasurements;
+      }
+
+      const kind: MeasurementKind = 'gclcms';
+
+      const times = reader.getDataVariable('scan_acquisition_time');
+      const tics = reader.getDataVariable('total_intensity');
+      const pointCount = reader.getDataVariable('point_count');
+      const massValues = reader.getDataVariable('mass_values');
+      const intensityValues = reader.getDataVariable('intensity_values');
+      addMeta(reader, reader.globalAttributes);
+
+      let index = 0;
+      const allMasses: Float64Array[] = [];
+      const allIntensities: Float64Array[] = [];
+      for (let size of pointCount) {
+        // Taken from: https://github.com/cheminfo/netcdf-gcms
+        const masses = new Float64Array(size);
+        const intensities = new Float64Array(size);
+        for (let j = 0; j < size; j++) {
+          masses[j] = massValues[index];
+          intensities[j] = intensityValues[index++];
+        }
+        allIntensities.push(intensities);
+        allMasses.push(masses);
+      }
+      newMeasurements[kind].entries.push({
+        id: v4(),
+        meta: reader.header.meta,
+        filename: file.name,
+        path: file.relativePath || '',
+        info: {},
+        title: reader.getAttribute('experiment_title'),
+        data: normalizeChromatogram(times, allMasses, allIntensities, tics),
+      });
+    }
+  }
+  return newMeasurements;
+};
+
+function normalizeChromatogram(time, masses, intensities, tics) {
+  let data: any = [];
+  for (let i = 0; i < time.length; i++) {
+    data.push({
+      meta: {},
+      info: {
+        time: { value: time[i], units: 's' },
+        tic: tics[i],
+      },
+      variables: {
+        x: {
+          symbol: 'X',
+          label: 'm/z',
+          units: '',
+          data: masses[i] || [],
+        },
+        y: {
+          symbol: 'Y',
+          label: 'relative intensity',
+          units: '',
+          data: intensities[i] || [],
+        },
+      },
+    });
+  }
+  return data;
+}
+
+function addMeta(reader, globalAttributes) {
+  reader.header.meta = {};
+  for (const item of globalAttributes) {
+    reader.header.meta[item.name] = item.value;
+  }
+}

src/app/data/loaders/jcampLoader.ts

@@ -11,7 +11,7 @@ import {
 
 export const jcampLoader: Loader = async function jcampLoader(
   fileCollection: FileCollection,
-) {
+): Promise<Measurements> {
   const newMeasurements: Measurements = getEmptyMeasurements();
 
   for (const file of fileCollection) {

src/app/data/loaders/proprietary/agilent/cary500Loader.ts

@@ -5,7 +5,7 @@ import { Loader, Measurements, getEmptyMeasurements } from '../../../DataState';
 
 export const cary500Loader: Loader = async function cary500Loader(
   fileCollection: FileCollection,
-) {
+): Promise<Measurements> {
   const newMeasurements: Measurements = getEmptyMeasurements();
 
   for (const file of fileCollection) {

src/app/data/loaders/spcLoader.ts

@@ -2,12 +2,12 @@ import { v4 } from '@lukeed/uuid';
 import type { FileCollection } from 'filelist-utils';
 import { parse, guessSpectraType } from 'spc-parser';
 
-import { getEmptyMeasurements, Loader } from '../DataState';
+import { getEmptyMeasurements, Loader, Measurements } from '../DataState';
 import type { MeasurementBase } from '../MeasurementBase';
 
 export const spcLoader: Loader = async function spcLoader(
   fileCollection: FileCollection,
-) {
+): Promise<Measurements> {
   let measurements = getEmptyMeasurements();
   for (const file of fileCollection) {
     if (file.name.match(/\.spc$/i)) {

src/app/data/loaders/wdfLoader.ts

@@ -6,7 +6,7 @@ import { getEmptyMeasurements, Loader, Measurements } from '../DataState';
 
 export const wdfLoader: Loader = async function wdfLoader(
   fileCollection: FileCollection,
-) {
+): Promise<Measurements> {
   const measurements: Measurements = getEmptyMeasurements();
 
   for (const file of fileCollection) {

src/app/main.tsx

@@ -19,6 +19,7 @@ createRoot(document.getElementById('root') as HTMLDivElement).render(
           paddingBlock: 5,
         }}
       >
+        <a href={import.meta.env.BASE_URL}>Blank</a>
         <a
           href={`${
             import.meta.env.BASE_URL
@@ -40,6 +41,20 @@ createRoot(document.getElementById('root') as HTMLDivElement).render(
         >
           UV-vis
         </a>
+        <a
+          href={`${
+            import.meta.env.BASE_URL
+          }?filelist=https://zakodium-oss.github.io/analysis-dataset/hplc.json`}
+        >
+          HPLC
+        </a>
+        <a
+          href={`${
+            import.meta.env.BASE_URL
+          }?filelist=https://zakodium-oss.github.io/analysis-dataset/gcms.json`}
+        >
+          GC/MS
+        </a>
         <a
           href={`${
             import.meta.env.BASE_URL