@sjcrh/proteinpaint-server 2.27.0 → 2.27.2

package/cards/databrowser.json

@@ -1,5 +1,5 @@
 {
-  "intro": "<b>We're excited to announce our newest app!</b> By uploading a dictionary in the Data Browser, one can explore the structure of their custom data. <br><br>Options to upload sample annotation and other data types are in development.", 
+  "intro": "By uploading a dictionary in the Data Browser, one can explore the structure of their custom data. <br><br>Options to upload sample annotation and other data types are in development.", 
   "ppcalls":[
     {
         "isUi": true,

package/cards/disco.json

@@ -0,0 +1,69 @@
+{
+    "ribbonMessage": "We're excited to announce our newest app, the <b>Disco Plot!</b>",
+    "ppcalls": [
+        {
+            "isUi": true,
+            "runargs": {
+                "noheader": true,
+                "nobox": 1,
+                "parseurl": false,
+                "tkui": "disco"
+            }
+        },
+        {
+            "message": "In this plot, all chromosomes of the genome are arranged in a circle. Multiple rings along the circle indicates example data points of mutation, copy number change, and structural variation/fusion events. Gene labels are based on mutation and sv/fusion events.",
+            "runargs": {
+                "noheader": true,
+                "nobox": true,
+                "genome": "hg38",
+                "disco": {
+                    "mlst": [
+                        {
+                            "alt": "T",
+                            "chr": "chr1",
+                            "class": "M",
+                            "dt": 1,
+                            "gene": "H3F3A",
+                            "isoform": "NM_002107",
+                            "mname": "K28M",
+                            "position": 226252135,
+                            "ref": "A"
+                        },
+                        {
+                            "dt": 4,
+                            "chr": "chr1",
+                            "start": 1,
+                            "stop": 100000000,
+                            "value": 0.5
+                        },
+                        {
+                            "dt": 4,
+                            "chr": "chr1",
+                            "start": 100000000,
+                            "stop": 200000000,
+                            "value": -0.5
+                        },
+                        {
+                            "chrA": "chr6",
+                            "posA": 3067605,
+                            "geneA": "MDC1",
+                            "chrB": "chr12",
+                            "posB": 61521661,
+                            "geneB": "KMT2D",
+                            "dt": 2,
+                            "strandA": "+",
+                            "strandB": "-"
+                        }
+                    ]
+                }
+            },
+            "testSpec": {
+                "button": 1,
+                "expected": {
+                    "svg": 1,
+                    "g": 2
+                }
+            }
+        }
+    ]
+}

package/cards/hic.json

@@ -1,4 +1,5 @@
 {
+    "ribbonMessage": "Versions 7, 8 and 9 Hi-C files are accepted.",
     "ppcalls": [
         {
             "label": "Hi-C",

package/cards/index.json

@@ -106,6 +106,7 @@
             "section": "tracks",
             "description": "Chromatin interaction at a locus",
             "image": "https://proteinpaint.stjude.org/ppdemo/images/hic-square.png",
+            "ribbon": { "text": "updated", "expireDate": "2023-10-31" },
             "sandboxJson": "hic",
             "searchterms": ["hic", "hicstraw", "chromatin interaction", "conformation"]
         },
@@ -285,7 +286,6 @@
             "section": "apps",
             "description": "2D scatter plot of samples & metadata",
             "image": "https://proteinpaint.stjude.org/ppdemo/images/scatterplot-square.png",
-            "ribbon": { "text": "updated", "expireDate": "2023-03-01" },
             "sandboxJson": "scatterplot",
             "searchterms": ["tSNE", "lasso"]
         },
@@ -307,6 +307,16 @@
             "ribbon": {"text":"beta"},
             "sandboxJson": "databrowser",
             "searchterms": ["clinical", "dictionary"]
+        },
+        {
+            "type": "card",
+            "name": "Disco Plot",
+            "section": "apps",
+            "description": "Circos-like plot of genome-wide mutational events",
+            "image": "https://proteinpaint.stjude.org/ppdemo/images/disco-square.png",
+            "ribbon": { "text":"new", "expireDate": "2023-10-31" },
+            "sandboxJson": "disco",
+            "searchterms": ["disco"]
         }
     ]
-}
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sjcrh/proteinpaint-server",
-  "version": "2.27.0",
+  "version": "2.27.2",
   "description": "a genomics visualization tool for exploring a cohort's genotype and phenotype data",
   "main": "server.js",
   "bin": "start.js",
@@ -71,6 +71,7 @@
     "jsonwebtoken": "^9.0.0",
     "jstat": "^1.9.3",
     "lazy": "^1.0.11",
+    "micromatch": "^4.0.5",
     "minimatch": "^3.1.2",
     "node-fetch": "^2.6.1",
     "partjson": "^0.58.1",

package/routes/{gdcMaf.ts → gdc.maf.ts}

@@ -1,4 +1,4 @@
-import { GdcMafResponse, File } from '#shared/types/routes/gdcMaf.ts'
+import { GdcMafResponse, File } from '#shared/types/routes/gdc.maf.ts'
 import { fileSize } from '#shared/fileSize.js'
 import path from 'path'
 import got from 'got'
@@ -6,7 +6,7 @@ import got from 'got'
 const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov'
 
 export const api = {
-	endpoint: 'gdcMaf',
+	endpoint: 'gdc/maf',
 	methods: {
 		get: {
 			init({ genomes }) {

package/routes/gdc.topMutatedGenes.ts

@@ -0,0 +1,130 @@
+import { GdcTopMutatedGeneResponse } from '#shared/types/routes/gdc.topMutatedGenes.ts'
+import path from 'path'
+import got from 'got'
+
+const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov'
+
+export const api = {
+	endpoint: 'gdc/topMutatedGenes',
+	methods: {
+		get: {
+			init({ genomes }) {
+				/*
+				genomes parameter is currently not used
+				could be used later to:
+				- verify hg38/GDC is on this instance and otherwise disable this route..
+				- perform conversion on gene name/id for future on needs
+				*/
+
+				return async (req: any, res: any): Promise<void> => {
+					try {
+						const genes = await getGenes(req.query)
+						const payload = { genes } as GdcTopMutatedGeneResponse
+						res.send(payload)
+					} catch (e: any) {
+						res.send({ status: 'error', error: e.message || e })
+					}
+				}
+			},
+			request: {
+				typeId: null
+				//valid: default to type checker
+			},
+			response: {
+				typeId: 'GdcTopMutatedGeneResponse'
+				// will combine this with type checker
+				//valid: (t) => {}
+			}
+		}
+	}
+}
+
+/*
+req.query {
+	filter0 // optional gdc GFF cohort filter, invisible and read only
+		FIXME should there be pp filter too?
+	geneFilter?: str
+	maxGenes: int
+}
+
+mayAddCGC2filter() are copied to
+/utils/gdc/topSSMgenes.js
+and hosted on https://proteinpaint.stjude.org/GDC/
+*/
+async function getGenes(q: any) {
+	let _f = { op: 'and', content: [] } // allow blank filter to test geneset edit ui (without filter)
+	if (q.filter0) {
+		if (typeof q.filter0 != 'object') throw 'filter0 not object'
+		_f = q.filter0
+	}
+	const response = await got.post(path.join(apihost, '/analysis/top_mutated_genes_by_project'), {
+		headers: { 'Content-Type': 'application/json', Accept: 'application/json' },
+		body: JSON.stringify({
+			size: q.maxGenes || 50,
+			fields: 'symbol',
+			filters: mayAddCGC2filter(_f, q.geneFilter)
+		})
+	})
+	const re = JSON.parse(response.body)
+	const genes = [] as string[]
+	for (const hit of re.data.hits) {
+		if (!hit.symbol) continue
+		genes.push(hit.symbol)
+	}
+	return genes
+}
+
+/*
+str:
+	stringified gdc filter object, should not include the "genes.is_cancer_gene_census" filter element
+geneFilter: str
+	if = 'CGC', insert following element into the filter and return stringified obj
+
+	{
+		"op":"and",
+		"content":[
+			{
+				"content":{ "field":"genes.is_cancer_gene_census", "value":["true"] },
+				"op":"in"
+			}
+		]
+	}
+*/
+function mayAddCGC2filter(f: any, geneFilter?: string) {
+	// reformulate f into f2
+	// f may be "in" or "and". f2 is always "and", in order to join in additional filters
+	let f2
+
+	if (f.op == 'in') {
+		// wrap f into f2
+		f2 = { op: 'and', content: [f] }
+	} else if (f.op == 'and') {
+		// no need to wrap
+		f2 = f
+	} else {
+		throw 'f.op not "in" or "and"'
+	}
+
+	// per Phil on 12/16/2022, following filters ensure to return IDH1 as 1st gene for gliomas
+	f2.content.push({
+		op: 'NOT',
+		content: {
+			field: 'ssms.consequence.transcript.annotation.vep_impact',
+			value: 'missing'
+		}
+	})
+	f2.content.push({
+		op: 'in',
+		content: {
+			field: 'ssms.consequence.transcript.consequence_type',
+			value: ['missense_variant', 'frameshift_variant', 'start_lost', 'stop_lost', 'stop_gained']
+		}
+	})
+
+	if (geneFilter == 'CGC') {
+		// using CGC genes, add in filter
+		f2.content.push({ op: 'in', content: { field: 'genes.is_cancer_gene_census', value: ['true'] } })
+	}
+
+	return f2
+}

package/routes/gdc.topVariablyExpressedGenes.ts

@@ -0,0 +1,97 @@
+import { GdcTopVariablyExpressedGenesResponse } from '#shared/types/routes/gdc.topVariablyExpressedGenes.ts'
+import { getCasesWithExressionDataFromCohort } from '../src/mds3.gdc.js'
+//import path from 'path'
+import got from 'got'
+import serverconfig from '../src/serverconfig.js'
+
+// TODO change when api is released to prod
+//const apihost = process.env.PP_GDC_HOST || 'https://api.gdc.cancer.gov'
+const apihost = 'https://uat-portal.gdc.cancer.gov/auth/api/v0/gene_expression/gene_selection'
+
+const gdcGenome = 'hg38'
+const gdcDslabel = 'GDC'
+
+export const api = {
+	endpoint: 'gdc/topVariablyExpressedGenes',
+	methods: {
+		get: {
+			init({ genomes }) {
+				return async (req: any, res: any): Promise<void> => {
+					try {
+						// following logic requires hg38 gdc dataset
+						const genome = genomes[gdcGenome]
+						if (!genome) throw 'hg38 genome missing'
+						const ds = genome.datasets?.[gdcDslabel]
+						if (!ds) throw 'gdc dataset missing'
+						const genes = await getGenes(req.query, ds)
+						const payload = { genes } as GdcTopVariablyExpressedGenesResponse
+						res.send(payload)
+					} catch (e: any) {
+						res.send({ status: 'error', error: e.message || e })
+					}
+				}
+			},
+			request: {
+				typeId: null
+				//valid: default to type checker
+			},
+			response: {
+				typeId: 'GdcTopVariablyExpressedGenesResponse'
+				// will combine this with type checker
+				//valid: (t) => {}
+			}
+		}
+	}
+}
+
+/*
+req.query {
+	filter0 // optional gdc GFF cohort filter, invisible and read only
+		FIXME should there be pp filter too?
+	maxGenes: int
+}
+
+ds { } // server-side ds object
+
+*/
+async function getGenes(q: any, ds: any) {
+	if (serverconfig.features.gdcGenes) {
+		// for testing only; delete when api issue is resolved
+		return serverconfig.features.gdcGenes as string[]
+	}
+
+	// based on current cohort, get list of cases with exp data, as input of next api query
+	const caseLst = await getCasesWithExressionDataFromCohort(q, ds)
+	if (caseLst.length == 0) {
+		// there are no cases with gene exp data
+		return [] as string[]
+	}
+
+	// change to this when api is available on prod
+	// const url = path.join(apihost, '/gene_expression/gene_selection')
+
+	const response = await got.post(apihost, {
+		headers: { 'Content-Type': 'application/json', Accept: 'application/json' },
+		body: JSON.stringify({
+			case_ids: caseLst,
+			//gene_ids: [] // this should not be a required parameter
+			size: q.maxGenes || 100
+		})
+	})
+
+	const re = JSON.parse(response.body)
+	// {"gene_selection":[{"gene_id":"ENSG00000141510","log2_uqfpkm_median":3.103430497010492,"log2_uqfpkm_stddev":0.8692021350485105,"symbol":"TP53"}, ... ]}
+
+	const genes = [] as string[]
+	if (!Array.isArray(re.gene_selection)) throw 're.gene_selection[] is not array'
+	for (const i of re.gene_selection) {
+		if (i.gene_id && typeof i.gene_id == 'string') {
+			genes.push(i.gene_id) // ensg
+		} 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
+}