@sjcrh/proteinpaint-shared 2.186.0 → 2.188.0

package/src/termdb.initbinconfig.js

@@ -1,130 +0,0 @@
-/*
-Initialize a bin configuration for a numeric dataset
-<data>: array of numeric data values
-<opts> (optional): object of options
-    {}: output bin config as JavaScript object (default)
-    {format: 'string'}: output bin config as JSON string
-*/
-export default function initBinConfig(data, opts = {}) {
-	if (!data.length)
-		return {
-			mode: "discrete",
-			type: "regular-bin",
-			startinclusive: true,
-			bin_size: null,
-			first_bin: { stop: null },
-		}
-	if (data.find((d) => !Number.isFinite(d)))
-		throw new Error("non-numeric values found")
-
-	let binConfig
-	const s = new Set(data)
-	if (s.size === 1) {
-		// single unique value in data array
-		// prepare custom bin config for 3 bins: first bin
-		// for values less than the value, second bin for values
-		// equal to the value, and third bin one for values
-		// greater than the value
-		// all data values will fall into the second bin
-		const value = [...s][0]
-		binConfig = {
-			type: "custom-bin",
-			lst: [
-				{
-					stop: value,
-					stopinclusive: false,
-					startunbounded: true,
-					label: "<" + value,
-				},
-				{
-					start: value,
-					stop: value,
-					startinclusive: true,
-					stopinclusive: true,
-					label: "=" + value,
-				},
-				{
-					start: value,
-					startinclusive: false,
-					stopunbounded: true,
-					label: ">" + value,
-				},
-			],
-		}
-	} else {
-		// multiple unique values in data array
-		// prepare regular bin config
-
-		// compute the bin size for a maximum bin number of 8
-		data.sort((a, b) => a - b)
-		const l = data.length
-		const min = data[0]
-		const max = data[l - 1]
-		const p5idx = Math.ceil(l * 0.05) - 1
-		const p98idx = Math.ceil(l * 0.98) - 1
-		const p5 = data[p5idx]
-		const p98 = data[p98idx]
-		// use 98th and 5th percentiles to compute bin size to reduce outlier influence
-		// if 98th = 5th, use max and min instead
-		const binSize = p98 != p5 ? (p98 - p5) / 8 : (max - min) / 8
-		// first bin stop will equal either (minimum + bin size) or (5th percentile), whichever is larger.
-		const firstBinStop = Math.max(min + binSize, p5)
-		// round the bin values
-		let [binSize_rnd, firstBinStop_rnd, lastBinStart_rnd, rounding] =
-			roundBinVals(binSize, firstBinStop, max, min)
-		// generate the bin configuration
-		binConfig = {
-			type: "regular-bin",
-			startinclusive: true,
-			bin_size: binSize_rnd,
-			first_bin: { stop: firstBinStop_rnd },
-		}
-		if (lastBinStart_rnd) binConfig.last_bin = { start: lastBinStart_rnd }
-		if (rounding) binConfig.rounding = rounding
-	}
-	if ("format" in opts) {
-		if (opts.format === "string") {
-			return JSON.stringify(binConfig)
-		} else {
-			throw "options are not in the correct format"
-		}
-	} else {
-		return binConfig
-	}
-}
-
-function roundBinVals(binSize, firstBinStop, max, min) {
-	let binSize_rnd, firstBinStop_rnd, lastBinStart_rnd, rounding
-	const log = Math.floor(Math.log10(binSize))
-	if (binSize >= 0.1 && binSize <= 2) {
-		// Round to the nearest one for small bin sizes
-		binSize_rnd = Math.round(binSize / (1 * 10 ** log)) * (1 * 10 ** log)
-		firstBinStop_rnd =
-			Math.round(firstBinStop / (1 * 10 ** log)) * (1 * 10 ** log)
-	} else {
-		// Round to the nearest five for large bin sizes
-		binSize_rnd = Math.round(binSize / (5 * 10 ** log)) * (5 * 10 ** log)
-		firstBinStop_rnd =
-			Math.round(firstBinStop / (5 * 10 ** log)) * (5 * 10 ** log)
-		if (binSize_rnd === 0) binSize_rnd = 1 * 10 ** log
-		if (firstBinStop_rnd === 0) firstBinStop_rnd = 1 * 10 ** log
-		if (binSize_rnd === 5 * 10 ** log && firstBinStop_rnd === 1 * 10 ** log)
-			firstBinStop_rnd = 5 * 10 ** log
-	}
-	if (firstBinStop_rnd < min) firstBinStop_rnd = firstBinStop_rnd * 2
-	// if the number of bins is above 8 after rounding, then set the last bin start to restrict the number of bins to 8
-	const eighthBinStop_rnd = firstBinStop_rnd + binSize_rnd * 7
-	if (max > eighthBinStop_rnd) {
-		lastBinStart_rnd = firstBinStop_rnd + binSize_rnd * 6
-	}
-	if (binSize < 1) {
-		const digits = Math.abs(log)
-		binSize_rnd = Number(binSize_rnd.toFixed(digits))
-		firstBinStop_rnd = Number(firstBinStop_rnd.toFixed(digits))
-		if (lastBinStart_rnd)
-			lastBinStart_rnd = Number(lastBinStart_rnd.toFixed(digits))
-		rounding = "." + digits + "f"
-	}
-	if (Object.is(firstBinStop_rnd, -0)) firstBinStop_rnd = 0
-	return [binSize_rnd, firstBinStop_rnd, lastBinStart_rnd, rounding]
-}

package/src/termdb.usecase.js

@@ -1,317 +0,0 @@
-import {
-	TermTypes,
-	isNumericTerm,
-	SINGLECELL_CELLTYPE,
-	SINGLECELL_GENE_EXPRESSION,
-	ISOFORM_EXPRESSION,
-} from "./terms.js"
-
-export const graphableTypes = new Set([
-	"categorical",
-	"integer",
-	"float",
-	"condition",
-	"survival",
-	"snplst",
-	"snplocus",
-	"geneVariant",
-	"samplelst",
-	"geneExpression",
-	ISOFORM_EXPRESSION,
-	"dtcnv",
-	"dtsnvindel",
-	"dtfusion",
-	"dtsv",
-	"date",
-	TermTypes.SSGSEA,
-	TermTypes.DNA_METHYLATION,
-	TermTypes.METABOLITE_INTENSITY,
-	TermTypes.PROTEOME_ABUNDANCE,
-	SINGLECELL_GENE_EXPRESSION,
-	SINGLECELL_CELLTYPE,
-	TermTypes.SNP,
-	TermTypes.TERM_COLLECTION,
-])
-
-/*
-isUsableTerm() will
-
-	- centralize the "allowed term" logic
-	which can be intricate or dataset-specific 
-	for certain terms or contexts
-	- make it easy to handle new term types
-
-Arguments:
-
-	term {}
-		.type: 'categorical', etc.
-		.child_types: []
-	
-	_usecase {}
-		.target (REQUIRED): 'barchart', 'regression', etc
-			- used as a switch-case "router" for additional use-specific logic
-			- other parameters, if applicable, are described in the route "handler" 
-		.detail 
-		  - a more specific detailed use case
-	
-
-	termdbConfig
-		optional. provides ds overrides on default rules via excludedTermtypeByTarget. for use on client
-
-	ds
-		optional. provides ds overrides when the function runs on backend
-		server-side dataset object that can supply overrides (in the form of functions) to the use case logic,
-		for example, to apply role-based allowed term uses or performance-related restrictions
-		to ancestor terms when a use case aggregates too many data points for a given chart type
-
-Returns
-
-	a Set{} with zero or more of the following strings:
-	- 'plot' if the term can be used in a plot chartType
-	- 'branch' if the term can be used only as an expandable tree branch, but not in a plot
-	- an empty Set means that the term has no valid uses, i.e, it cannot be used either for plotting or as a tree branch
-*/
-export function isUsableTerm(term, _usecase, termdbConfig, ds) {
-	const usecase = _usecase || {}
-
-	// may apply dataset specific override filter for a use case
-	if (typeof ds?.usecase?.[usecase.target] == "function") {
-		return ds.usecase[usecase.target](term, usecase)
-	}
-
-	// if (term.isprivate && !user.roleCanUse(term)) return false
-
-	const uses = new Set()
-	// note: expects term.child_types to be null if term.isleaf == true
-	const child_types = term.child_types || []
-	// default handling
-	switch (usecase.target) {
-		case "barchart":
-		case "violin":
-		case "boxplot":
-		case "summary":
-			if (term.type && term.type !== "survival") uses.add("plot")
-			if (hasAllowedChildTypes(child_types, ["survival"])) uses.add("branch")
-			return uses
-
-		case "summaryInput":
-			if (usecase.detail === "term2" || usecase.detail == "term0") {
-				if (term.type && term.type !== "survival") uses.add("plot")
-				if (hasAllowedChildTypes(child_types, ["survival"])) uses.add("branch")
-				return uses
-			} else {
-				if (graphableTypes.has(term.type)) uses.add("plot")
-				if (!term.isleaf) uses.add("branch")
-				return uses
-			}
-
-		case "matrix":
-			if (term.type) uses.add("plot")
-			if (!term.isleaf) uses.add("branch")
-			return uses
-
-		case "table":
-			if (usecase.detail == "term") uses.add("plot")
-			if (child_types.length > 1) uses.add("branch")
-			return uses
-
-		case "sampleScatter":
-			if (usecase.detail == "numeric") {
-				if (isNumericTerm(term)) {
-					uses.add("plot")
-				}
-				if (hasNumericChild(child_types)) uses.add("branch")
-			}
-			// Commenting out for now. May need later for another single
-			// cell term. Revisit logic at that time.
-			// else if (usecase?.specialCase?.type == 'singleCell') {
-			// 		if (term.type && term.type.startsWith('singleCell')) {
-			// 			if (term.plot && term.plot == usecase.specialCase?.config.name) {
-			// 				uses.add('plot')
-			// 			}
-			// 		}
-			// }
-			else {
-				if (graphableTypes.has(term.type)) uses.add("plot")
-				if (!term.isleaf) uses.add("branch")
-			}
-			return uses
-		case "runChart2":
-			if (usecase.detail == "date" || usecase.detail == "xtw") {
-				if (term.type == "date") {
-					uses.add("plot")
-				}
-				if (child_types.includes("date")) uses.add("branch")
-			} else if (usecase.detail == "numeric") {
-				if (isNumericTerm(term) && term.type != "date") {
-					uses.add("plot")
-				}
-				if (hasNumericChild(child_types)) uses.add("branch")
-			} else {
-				if (graphableTypes.has(term.type)) uses.add("plot")
-				if (!term.isleaf) uses.add("branch")
-			}
-			return uses
-
-		case "numericDictTermCluster":
-			if (!usecase.detail?.exclude?.includes(term.id)) {
-				if (isNumericTerm(term)) {
-					uses.add("plot")
-				}
-				if (hasNumericChild(child_types)) {
-					uses.add("branch")
-				}
-			}
-			return uses
-
-		case "termCollections":
-			if (usecase.detail?.termIds?.includes(term.id)) uses.add("plot")
-			if (usecase.detail?.branchIds?.includes(term.id)) uses.add("branch")
-			return uses
-
-		case "profileForms":
-			if (!term.isleaf) {
-				const ancestors = term.id.split("__").length //depends on using the __ naming convension!
-				if (ancestors == 3) {
-					// 3rd level term is a domain, we show the templates associated to this domain
-					uses.add("plot")
-				} else if (ancestors < 3) uses.add("branch")
-			}
-			return uses
-
-		// case 'boxplot':
-		// 	if (term.type == 'float' || term.type == 'integer') uses.add('plot')
-		// 	if (usecase.detail === 'term2' && hasNumericChild(child_types)) uses.add('branch')
-		// 	return uses
-
-		case "cuminc":
-			if (usecase.detail == "term") {
-				if (term.type == "condition") uses.add("plot")
-				if (child_types.includes("condition")) uses.add("branch")
-				return uses
-			}
-			if (usecase.detail === "term2" || usecase.detail == "term0") {
-				if (term.type && term.type != "condition" && term.type != "survival")
-					uses.add("plot")
-				if (hasAllowedChildTypes(child_types, ["condition", "survival"]))
-					uses.add("branch")
-				return uses
-			}
-			return uses
-
-		case "survival":
-			if (usecase.detail == "term") {
-				if (term.type == "survival") uses.add("plot")
-				if (child_types.includes("survival")) uses.add("branch")
-				return uses
-			}
-			if (usecase.detail === "term2" || usecase.detail == "term0") {
-				if (term.type && term.type != "survival") uses.add("plot")
-				if (hasAllowedChildTypes(child_types, ["survival"])) uses.add("branch")
-				return uses
-			}
-			return uses
-
-		case "regression":
-			if (usecase.detail == "outcome") {
-				if (usecase.regressionType == "linear") {
-					if (term.type == "float" || term.type == "integer") uses.add("plot")
-					if (hasNumericChild(child_types)) uses.add("branch")
-					return uses
-				}
-				if (usecase.regressionType == "logistic") {
-					if (term.type && term.type != "survival") uses.add("plot")
-					if (hasAllowedChildTypes(child_types, ["survival"]))
-						uses.add("branch")
-					return uses
-				} else if (usecase.regressionType == "cox") {
-					if (term.type == "condition" || term.type == "survival")
-						uses.add("plot")
-					if (
-						child_types.includes("condition") ||
-						child_types.includes("survival")
-					)
-						uses.add("branch")
-					return uses
-				}
-			}
-
-			if (usecase.detail == "independent") {
-				if (
-					term.type == "float" ||
-					term.type == "integer" ||
-					term.type == "categorical" ||
-					term.type == "samplelst"
-				)
-					uses.add("plot")
-				if (hasChildTypes(child_types, ["categorical", "float", "integer"]))
-					uses.add("branch")
-				return uses
-			}
-			return uses
-
-		case "filter": {
-			// apply "exlst" to other targets as needed
-			const exlst = termdbConfig?.excludedTermtypeByTarget?.filter
-			if (exlst) {
-				if (graphableTypes.has(term.type) && !exlst.includes(term.type))
-					uses.add("plot")
-				if (child_types.find((t) => !exlst.includes(t))) uses.add("branch") // there's a non-excluded child type, allow branch to show
-				return uses
-			}
-			// no specific rule for filter. use default rules
-			if (graphableTypes.has(term.type)) uses.add("plot")
-			if (!term.isleaf) uses.add("branch")
-			return uses
-		}
-
-		case "correlationVolcano":
-			if (usecase.detail == "numeric") {
-				if (isNumericTerm(term)) {
-					uses.add("plot")
-				}
-				if (hasNumericChild(child_types)) uses.add("branch")
-			} else {
-				if (graphableTypes.has(term.type)) uses.add("plot")
-				if (!term.isleaf) uses.add("branch")
-			}
-			return uses
-
-		case "proteinView":
-			if (term.type == TermTypes.PROTEOME_ABUNDANCE) uses.add("plot")
-			if (child_types.includes(TermTypes.PROTEOME_ABUNDANCE)) uses.add("branch")
-			return uses
-
-		default:
-			if (graphableTypes.has(term.type)) uses.add("plot")
-			if (!term.isleaf) uses.add("branch")
-			return uses
-	}
-}
-
-// determine if the term has at least one child type that
-// is not excluded
-function hasAllowedChildTypes(child_types, excluded_types) {
-	if (!child_types.length) {
-		// term does not have children
-		return false
-	}
-	if (!excluded_types?.length) {
-		// no excluded types
-		return true
-	}
-	if (child_types.some((type) => !excluded_types.includes(type))) {
-		// at least one child type is not excluded
-		return true
-	}
-}
-
-function hasNumericChild(child_types) {
-	return child_types.includes("float") || child_types.includes("integer")
-}
-
-function hasChildTypes(child_types, expected_types) {
-	for (const a of expected_types) {
-		if (child_types.includes(a)) return true
-	}
-}