@sjcrh/proteinpaint-server 2.39.0 → 2.39.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.39.0",
+  "version": "2.39.1",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "server.js",
   "bin": "start.js",

package/routes/gdc.maf.ts

@@ -3,15 +3,28 @@ import path from 'path'
 import got from 'got'
 import serverconfig from '#src/serverconfig.js'
 
-const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov'
-const maxFileNumber = 1000
+/*
+this route lists available gdc MAF files based on user's cohort filter
+and return them to client to be shown in a table for selection
+*/
+
+const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov' // to switch to serverconfig export
+
+const maxFileNumber = 1000 // determines max number of files to return to client
+// preliminary testing:
+// 36s for 1000 (87Mb)
+// 78s for 2000 (177Mb)
+// if safe to increase to 2000, maybe fast when this runs in gdc env
+
 const allowedWorkflowType = 'Aliquot Ensemble Somatic Variant Merging and Masking'
-const maxTotalSizeCompressed = serverconfig.features.gdcMafMaxFileSize || 50000000 // 50Mb
+
+// change to 400 so it won't limit number of files; should keep this setting as a safeguard; also it's fast to check file size (.5s in gdc.mafBuild.ts)
+export const maxTotalSizeCompressed = serverconfig.features.gdcMafMaxFileSize || 400000000 // 400Mb
 
 export const api = {
 	endpoint: 'gdc/maf',
 	methods: {
-		get: {
+		all: {
 			init,
 			request: {
 				typeId: 'GdcMafRequest'

package/routes/gdc.mafBuild.ts

@@ -5,9 +5,9 @@ import { run_rust_stream } from '@sjcrh/proteinpaint-rust'
 import serverconfig from '#src/serverconfig.js'
 import Readable from 'stream'
 import { GdcMafBuildRequest } from '#shared/types/routes/gdc.mafBuild.ts'
+import { maxTotalSizeCompressed } from './gdc.maf.ts'
 
 const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov'
-const maxTotalSizeCompressed = serverconfig.features.gdcMafMaxFileSize || 50000000 // 50Mb
 
 export const api = {
 	endpoint: 'gdc/mafBuild',
@@ -43,6 +43,7 @@ async function buildMaf(q: GdcMafBuildRequest, res: any) {
 	const t0 = Date.now()
 
 	const fileLst2 = (await getFileLstUnderSizeLimit(q.fileIdLst)) as string[]
+
 	if (serverconfig.debugmode)
 		console.log(
 			`${fileLst2.length} out of ${q.fileIdLst.length} input MAF files accepted by size limit`,
@@ -60,11 +61,12 @@ async function buildMaf(q: GdcMafBuildRequest, res: any) {
 	res.setHeader('Content-Disposition', 'attachment; filename=cohort.maf.gz')
 	rustStream.pipe(res)
 
-	if (serverconfig.debugmode) console.log('rust gdcmaf', Date.now() - t0)
-
 	rustStream.on('end', () => {
+		// report amount of time taken to run rust
+		if (serverconfig.debugmode) console.log('rust gdcmaf', Date.now() - t0)
 		res.end()
 	})
+
 	rustStream.on('error', err => {
 		console.error(err)
 		res.statusCode = 500
@@ -85,7 +87,7 @@ async function getFileLstUnderSizeLimit(lst: string[]) {
 			op: 'in',
 			content: { field: 'file_id', value: lst }
 		},
-		size: 1000,
+		size: 10000,
 		fields: 'file_size'
 	}
 	const headers = { 'Content-Type': 'application/json', Accept: 'application/json' }

package/routes/hicdata.ts

@@ -40,7 +40,7 @@ export const api: any = {
 	}
 }
 
-function init({ genomes }) {
+function init() {
 	return async (req: any, res: any): Promise<void> => {
 		try {
 			const payload = await handle_hicdata(req.query as HicdataRequest)
@@ -60,7 +60,7 @@ function handle_hicdata(q: HicdataRequest) {
 		if (e) reject({ error: 'illegal file name' })
 
 		const par = [
-			// TODO add option for observed/oe
+			q.matrixType || 'observed',
 			q.nmeth || 'NONE',
 			file,
 			q.pos1,

package/routes/termdb.cluster.ts

@@ -1,5 +1,4 @@
 import { TermdbClusterRequest, TermdbClusterResponse } from '#shared/types/routes/termdb.cluster.ts'
-import fs from 'fs'
 import path from 'path'
 import * as utils from '#src/utils.js'
 import serverconfig from '#src/serverconfig.js'
@@ -12,7 +11,7 @@ import { dtgeneexpression } from '#shared/common.js'
 export const api = {
 	endpoint: 'termdb/cluster',
 	methods: {
-		get: {
+		all: {
 			init,
 			request: {
 				typeId: 'TermdbClusterRequest'
@@ -20,10 +19,6 @@ export const api = {
 			response: {
 				typeId: 'TermdbClusterResponse'
 			}
-		},
-		post: {
-			alternativeFor: 'get',
-			init
 		}
 	}
 }
@@ -121,8 +116,8 @@ async function validateNative(q: GeneExpressionQueryNative, ds: any, genome: any
 	q.get = async (param: TermdbClusterRequest) => {
 		const limitSamples = await mayLimitSamples(param, q.samples, ds)
 		if (limitSamples?.size == 0) {
-			// got 0 sample after filtering, return blank array for no data
-			return new Set()
+			// got 0 sample after filtering, must still return expected structure with no data
+			return { gene2sample2value: new Set(), byTermId: {}, bySampleId: {} }
 		}
 
 		// has at least 1 sample passing filter and with exp data

package/routes/termdb.getdescrstats.ts

@@ -1,7 +1,6 @@
 import { getdescrstatsRequest, getdescrstatsResponse } from '#shared/types/routes/termdb.getdescrstats.ts'
-import roundValue from '#shared/roundValue.js'
-import computePercentile from '#shared/compute.percentile.js'
-import * as termdbsql from '#src/termdb.sql.js'
+import * as termdbsql from '../src/termdb.sql.js'
+import Summarystats from '../shared/descriptive.stats.js'
 
 export const api: any = {
 	endpoint: 'termdb/descrstats',
@@ -104,42 +103,5 @@ async function trigger_getdescrstats(q: any, res: any, ds: any) {
 		}
 		values.push(value)
 	}
-
-	// compute statistics
-	// total
-	const total = values.length
-
-	// mean
-	const sum = values.reduce((a, b) => a + b, 0)
-	const mean = sum / total
-
-	// percentiles
-	const p25 = computePercentile(values, 25)
-	const median = computePercentile(values, 50)
-	const p75 = computePercentile(values, 75)
-
-	// standard deviation
-	// get sum of squared differences from mean
-	const sumSqDiff = values.map(v => (v - mean) ** 2).reduce((a, b) => a + b, 0)
-	// get variance
-	const variance = sumSqDiff / (values.length - 1)
-	// get standard deviation
-	const sd = Math.sqrt(variance)
-
-	// min/max
-	const min = Math.min(...values)
-	const max = Math.max(...values)
-
-	res.send({
-		values: [
-			{ id: 'total', label: 'n', value: total },
-			{ id: 'min', label: 'Minimum', value: roundValue(min, 2) },
-			{ id: 'p25', label: '1st quartile', value: roundValue(p25, 2) },
-			{ id: 'median', label: 'Median', value: roundValue(median, 2) },
-			{ id: 'mean', label: 'Mean', value: roundValue(mean, 2) },
-			{ id: 'p75', label: '3rd quartile', value: roundValue(p75, 2) },
-			{ id: 'max', label: 'Maximum', value: roundValue(max, 2) },
-			{ id: 'sd', label: 'Standard deviation', value: roundValue(sd, 2) }
-		]
-	} as getdescrstatsResponse)
+	res.send(Summarystats(values) as getdescrstatsResponse)
 }

package/routes/termdb.getrootterm.ts

@@ -0,0 +1,67 @@
+import { getroottermRequest, getroottermResponse } from '#shared/types/routes/termdb.getrootterm.ts'
+
+export const api: any = {
+	endpoint: 'termdb/rootterm',
+	methods: {
+		get: {
+			init,
+			request: {
+				typeId: 'getroottermRequest'
+			},
+			response: {
+				typeId: 'getroottermResponse'
+			},
+			examples: [
+				{
+					request: {
+						body: {
+							genome: 'hg38-test',
+							dslabel: 'TermdbTest',
+							embedder: 'localhost',
+							default_rootterm: 1,
+							cohortValues: 'ABC'
+						}
+					},
+					response: {
+						header: { status: 200 }
+					}
+				}
+			]
+		},
+		post: {
+			alternativeFor: 'get',
+			init
+		}
+	}
+}
+
+function init({ genomes }) {
+	return async (req: any, res: any): Promise<void> => {
+		const q = req.query as getroottermRequest
+		try {
+			const g = genomes[req.query.genome]
+			if (!g) throw 'invalid genome name'
+			const ds = g.datasets[req.query.dslabel]
+			if (!ds) throw 'invalid dataset name'
+			const tdb = ds.cohort.termdb
+			if (!tdb) throw 'invalid termdb object'
+
+			await trigger_rootterm(q, res, tdb) // as getroottermResponse
+		} catch (e) {
+			// eslint-disable-next-line @typescript-eslint/ban-ts-comment
+			// @ts-ignore
+			res.send({ error: e?.message || e })
+			if (e instanceof Error && e.stack) console.log(e)
+		}
+	}
+}
+
+async function trigger_rootterm(
+	q: { cohortValues: any; treeFilter: any },
+	res: { send: (arg0: { lst: any }) => void },
+	tdb: { q: { getRootTerms: (arg0: any, arg1: any) => any } }
+) {
+	const cohortValues = q.cohortValues ? q.cohortValues : ''
+	const treeFilter = q.treeFilter ? q.treeFilter : ''
+	res.send({ lst: await tdb.q.getRootTerms(cohortValues, treeFilter) } as getroottermResponse)
+}

package/routes/termdb.gettermchildren.ts

@@ -0,0 +1,82 @@
+import { gettermchildrenRequest, gettermchildrenResponse } from '#shared/types/routes/termdb.gettermchildren.ts'
+import { copy_term } from '#src/termdb.js'
+
+export const api: any = {
+	endpoint: 'termdb/termchildren',
+	methods: {
+		get: {
+			init,
+			request: {
+				typeId: 'gettermchildrenRequest'
+			},
+			response: {
+				typeId: 'gettermchildrenResponse'
+			},
+			examples: [
+				{
+					request: {
+						body: {
+							genome: 'hg38-test',
+							dslabel: 'TermdbTest',
+							embedder: 'localhost',
+							get_children: 1,
+							cohortValues: 'ABC',
+							tid: 'GO:0000001'
+						}
+					},
+					response: {
+						header: { status: 200 }
+					}
+				}
+			]
+		},
+		post: {
+			alternativeFor: 'get',
+			init
+		}
+	}
+}
+
+function init({ genomes }) {
+	return async (req: any, res: any): Promise<void> => {
+		const q = req.query as gettermchildrenRequest
+		try {
+			const g = genomes[req.query.genome]
+			if (!g) throw 'invalid genome name'
+			const ds = g.datasets[req.query.dslabel]
+			if (!ds) throw 'invalid dataset name'
+			const tdb = ds.cohort.termdb
+			if (!tdb) throw 'invalid termdb object'
+
+			await trigger_children(q, res, tdb)
+		} catch (e) {
+			// eslint-disable-next-line @typescript-eslint/ban-ts-comment
+			// @ts-ignore
+			res.send({ error: e?.message || e })
+			if (e instanceof Error && e.stack) console.log(e)
+		}
+	}
+}
+
+async function trigger_children(
+	q: {
+		genome?: string
+		dslabel?: string
+		embedder?: string
+		get_children?: number
+		tid: any
+		cohortValues?: any
+		treeFilter?: any
+	},
+	res: { send: (arg0: gettermchildrenResponse) => void },
+	tdb: { q: { getTermChildren: (arg0: any, arg1: any, arg2: any) => any } }
+) {
+	/* get children terms
+may apply ssid: a premade sample set
+*/
+	if (!q.tid) throw 'no parent term id'
+	const cohortValues = q.cohortValues ? q.cohortValues : ''
+	const treeFilter = q.treeFilter ? q.treeFilter : ''
+	const terms = await tdb.q.getTermChildren(q.tid, cohortValues, treeFilter)
+	res.send({ lst: terms.map(copy_term) } as gettermchildrenResponse)
+}