feat: enable using custom assembly by specifying chrom sizes (#776)

* feat: support custom chrom sizes * chore: add a util function for creating chromsize URL
gosling-lang · Jul 24, 2022 · b08334a · b08334a
1 parent 4fa101d
commit b08334a
Show file tree

Hide file tree

Showing 14 changed files with 259 additions and 68 deletions.
diff --git a/editor/example/json-spec/sars-cov-2.ts b/editor/example/json-spec/sars-cov-2.ts
@@ -7,6 +7,7 @@ export const EX_TRACK_SARS_COV_2_GENES: OverlaidTracks = {
     data: {
         type: 'csv',
         url: 'https://s3.amazonaws.com/gosling-lang.org/data/COVID/NC_045512.2-Genes.csv',
+        chromosomeField: 'Accession',
         genomicFields: ['Start', 'Stop']
     },
     tracks: [
@@ -49,7 +50,7 @@ export const EX_TRACK_SARS_COV_2_GENES: OverlaidTracks = {
 export const EX_SPEC_SARS_COV_2: GoslingSpec = {
     title: 'SARS-CoV-2',
     subtitle: 'Data Source: WashU Virus Genome Browser, NCBI, GISAID',
-    assembly: 'unknown',
+    assembly: [['NC_045512.2', 29903]],
     layout: 'linear',
     spacing: 50,
     views: [
@@ -112,6 +113,7 @@ export const EX_SPEC_SARS_COV_2: GoslingSpec = {
                     data: {
                         type: 'csv',
                         url: 'https://s3.amazonaws.com/gosling-lang.org/data/COVID/sars-cov-2_Sprot_annot_sorted.bed',
+                        chromosomeField: 'Accession',
                         genomicFields: ['Start', 'Stop']
                     },
                     tracks: [
@@ -171,6 +173,7 @@ export const EX_SPEC_SARS_COV_2: GoslingSpec = {
                     data: {
                         type: 'csv',
                         url: 'https://s3.amazonaws.com/gosling-lang.org/data/COVID/TRS-L-dependent_recombinationEvents_sorted.bed',
+                        chromosomeField: 'Accession',
                         genomicFields: ['Start1', 'Stop1', 'Start2', 'Stop2'],
                         sampleLength: 100
                     },

diff --git a/schema/gosling.schema.json b/schema/gosling.schema.json
@@ -13,17 +13,43 @@
       "type": "string"
     },
     "Assembly": {
-      "enum": [
-        "hg38",
-        "hg19",
-        "hg18",
-        "hg17",
-        "hg16",
-        "mm10",
-        "mm9",
-        "unknown"
-      ],
-      "type": "string"
+      "anyOf": [
+        {
+          "const": "hg38",
+          "type": "string"
+        },
+        {
+          "const": "hg19",
+          "type": "string"
+        },
+        {
+          "const": "hg18",
+          "type": "string"
+        },
+        {
+          "const": "hg17",
+          "type": "string"
+        },
+        {
+          "const": "hg16",
+          "type": "string"
+        },
+        {
+          "const": "mm10",
+          "type": "string"
+        },
+        {
+          "const": "mm9",
+          "type": "string"
+        },
+        {
+          "const": "unknown",
+          "type": "string"
+        },
+        {
+          "$ref": "#/definitions/ChromSizes"
+        }
+      ]
     },
     "AxisPosition": {
       "enum": [
@@ -289,6 +315,23 @@
       ],
       "type": "object"
     },
+    "ChromSizes": {
+      "description": "Custom chromosome sizes, e.g., [[\"foo\", 1000], [\"bar\", 300], [\"baz\", 240]]",
+      "items": {
+        "items": [
+          {
+            "type": "string"
+          },
+          {
+            "type": "number"
+          }
+        ],
+        "maxItems": 2,
+        "minItems": 2,
+        "type": "array"
+      },
+      "type": "array"
+    },
     "Chromosome": {
       "enum": [
         "1",

diff --git a/schema/higlass.schema.json b/schema/higlass.schema.json
@@ -3,17 +3,60 @@
   "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Assembly": {
-      "enum": [
-        "hg38",
-        "hg19",
-        "hg18",
-        "hg17",
-        "hg16",
-        "mm10",
-        "mm9",
-        "unknown"
-      ],
-      "type": "string"
+      "anyOf": [
+        {
+          "const": "hg38",
+          "type": "string"
+        },
+        {
+          "const": "hg19",
+          "type": "string"
+        },
+        {
+          "const": "hg18",
+          "type": "string"
+        },
+        {
+          "const": "hg17",
+          "type": "string"
+        },
+        {
+          "const": "hg16",
+          "type": "string"
+        },
+        {
+          "const": "mm10",
+          "type": "string"
+        },
+        {
+          "const": "mm9",
+          "type": "string"
+        },
+        {
+          "const": "unknown",
+          "type": "string"
+        },
+        {
+          "$ref": "#/definitions/ChromSizes"
+        }
+      ]
+    },
+    "ChromSizes": {
+      "description": "Custom chromosome sizes, e.g., [[\"foo\", 1000], [\"bar\", 300], [\"baz\", 240]]",
+      "items": {
+        "items": [
+          {
+            "type": "string"
+          },
+          {
+            "type": "number"
+          }
+        ],
+        "maxItems": 2,
+        "minItems": 2,
+        "type": "array"
+      },
+      "type": "array"
     },
     "CombinedTrack": {
       "additionalProperties": false,

diff --git a/schema/template.schema.json b/schema/template.schema.json
@@ -13,17 +13,43 @@
       "type": "string"
     },
     "Assembly": {
-      "enum": [
-        "hg38",
-        "hg19",
-        "hg18",
-        "hg17",
-        "hg16",
-        "mm10",
-        "mm9",
-        "unknown"
-      ],
-      "type": "string"
+      "anyOf": [
+        {
+          "const": "hg38",
+          "type": "string"
+        },
+        {
+          "const": "hg19",
+          "type": "string"
+        },
+        {
+          "const": "hg18",
+          "type": "string"
+        },
+        {
+          "const": "hg17",
+          "type": "string"
+        },
+        {
+          "const": "hg16",
+          "type": "string"
+        },
+        {
+          "const": "mm10",
+          "type": "string"
+        },
+        {
+          "const": "mm9",
+          "type": "string"
+        },
+        {
+          "const": "unknown",
+          "type": "string"
+        },
+        {
+          "$ref": "#/definitions/ChromSizes"
+        }
+      ]
     },
     "AxisPosition": {
       "enum": [
@@ -443,6 +469,23 @@
         }
       ]
     },
+    "ChromSizes": {
+      "description": "Custom chromosome sizes, e.g., [[\"foo\", 1000], [\"bar\", 300], [\"baz\", 240]]",
+      "items": {
+        "items": [
+          {
+            "type": "string"
+          },
+          {
+            "type": "number"
+          }
+        ],
+        "maxItems": 2,
+        "minItems": 2,
+        "type": "array"
+      },
+      "type": "array"
+    },
     "Chromosome": {
       "enum": [
         "1",

diff --git a/scripts/setup-vitest.js b/scripts/setup-vitest.js
@@ -24,13 +24,15 @@ apis.forEach(api => {
     global.window[api] = canvasWindow[api];
 });
 
-// jsdom doesn't come with a WebCrypto implementation (required for uuid)
 beforeAll(() => {
+    // jsdom doesn't come with a WebCrypto implementation (required for uuid)
     global.crypto = {
         getRandomValues: function (buffer) {
             return randomFillSync(buffer);
         }
     };
+    // jsdom doesn't come with a `URL.createObjectURL` implementation
+    global.URL.createObjectURL = () => { return ''; };
 });
 
 afterAll(() => {

diff --git a/src/core/gosling.schema.ts b/src/core/gosling.schema.ts
@@ -86,7 +86,10 @@ export type ResponsiveSpecOfMultipleViews = {
 
 export type Layout = 'linear' | 'circular';
 export type Orientation = 'horizontal' | 'vertical';
-export type Assembly = 'hg38' | 'hg19' | 'hg18' | 'hg17' | 'hg16' | 'mm10' | 'mm9' | 'unknown';
+
+/** Custom chromosome sizes, e.g., [["foo", 1000], ["bar", 300], ["baz", 240]] */
+export type ChromSizes = [string, number][];
+export type Assembly = 'hg38' | 'hg19' | 'hg18' | 'hg17' | 'hg16' | 'mm10' | 'mm9' | 'unknown' | ChromSizes;
 export type ZoomLimits = [number | null, number | null];
 
 export interface CommonViewDef {

diff --git a/src/core/higlass-model.ts b/src/core/higlass-model.ts
@@ -12,7 +12,7 @@ import { insertItemToArray } from './utils/array';
 
 export const HIGLASS_AXIS_SIZE = 30;
 
-const getViewTemplate = (assembly?: string) => {
+const getViewTemplate = (assembly?: Assembly) => {
     return {
         genomePositionSearchBoxVisible: false,
         genomePositionSearchBox: {
@@ -74,7 +74,7 @@ export class HiGlassModel {
         return this;
     }
 
-    public addDefaultView(uid: string, assembly?: string) {
+    public addDefaultView(uid: string, assembly?: Assembly) {
         this.hg.views.push(JSON.parse(JSON.stringify({ ...getViewTemplate(assembly), uid })));
         return this;
     }

diff --git a/src/core/utils/assembly.test.ts b/src/core/utils/assembly.test.ts
@@ -2,14 +2,22 @@ import { getChromInterval, getChromTotalSize, getRelativeGenomicPosition, GET_CH
 import { CHROM_SIZE_HG38 } from './chrom-size';
 
 describe('Assembly', () => {
-    it('Random chromosome name', () => {
+    it('Missing chromosome name', () => {
         expect(GET_CHROM_SIZES().total).toBeDefined();
-        expect(GET_CHROM_SIZES('chrmosome1').total).toBeDefined();
-        expect(GET_CHROM_SIZES('random').total).toBeDefined();
     });
     it('hg38 ChromSizes', () => {
         expect(GET_CHROM_SIZES('hg38').size).toEqual(CHROM_SIZE_HG38);
     });
+    it('Custom ChromSizes', () => {
+        expect(
+            GET_CHROM_SIZES([
+                ['foo', 100],
+                ['bar', 1000]
+            ]).total
+        ).toEqual(1100);
+        // use default assembly instead when the size is zero
+        expect(GET_CHROM_SIZES([]).total).toEqual(GET_CHROM_SIZES().total);
+    });
     it('Chromosome total size calculation', () => {
         expect(getChromTotalSize({ 1: 1, 2: 999 })).toEqual(1000);
     });

diff --git a/src/core/utils/assembly.ts b/src/core/utils/assembly.ts
@@ -1,4 +1,4 @@
-import type { GenomicPosition } from '@gosling.schema';
+import type { Assembly, ChromSizes, GenomicPosition } from '@gosling.schema';
 import {
     CHROM_SIZE_HG16,
     CHROM_SIZE_HG17,
@@ -19,7 +19,7 @@ export interface ChromSize {
 /**
  * Get relative chromosome position (e.g., `100` => `{ chromosome: 'chr1', position: 100 }`)
  */
-export function getRelativeGenomicPosition(absPos: number, assembly?: string): GenomicPosition {
+export function getRelativeGenomicPosition(absPos: number, assembly?: Assembly): GenomicPosition {
     const [chromosome, absInterval] = Object.entries(GET_CHROM_SIZES(assembly).interval).find(d => {
         const [start, end] = d[1];
         return start <= absPos && absPos < end;
@@ -33,13 +33,32 @@ export function getRelativeGenomicPosition(absPos: number, assembly?: string): G
     return { chromosome, position: absPos - absInterval[0] };
 }
 
+/**
+ * Generate a URL for custom chrom sizes
+ * @param chromSizes A custom assembly that specifies chromosomes and their sizes
+ */
+function createChromSizesUrl(chromSizes: ChromSizes): string {
+    const text = chromSizes.map(d => d.join('\t')).join('\n');
+    const tsv = new Blob([text], { type: 'text/tsv' });
+    return URL.createObjectURL(tsv);
+}
+
 /**
  * Get chromosome sizes.
  * @param assembly (default: 'hg38')
  */
-export function GET_CHROM_SIZES(assembly?: string): ChromSize {
-    if (assembly && assembly in CRHOM_SIZES) {
+export function GET_CHROM_SIZES(assembly?: Assembly): ChromSize {
+    if (assembly && typeof assembly === 'string' && assembly in CRHOM_SIZES) {
         return CRHOM_SIZES[assembly];
+    } else if (Array.isArray(assembly) && assembly.length !== 0) {
+        // custom assembly
+        const size = Object.fromEntries(assembly);
+        return {
+            size,
+            interval: getChromInterval(size),
+            total: getChromTotalSize(size),
+            path: createChromSizesUrl(assembly)
+        };
     } else {
         // We do not have that assembly prepared, so return a default one.
         return CRHOM_SIZES.hg38;

diff --git a/src/core/utils/scales.ts b/src/core/utils/scales.ts
@@ -20,7 +20,7 @@ export function getNumericDomain(domain: Domain, assembly?: Assembly) {
         if (domain.chromosome.includes('chr')) {
             domain.chromosome = domain.chromosome.replace('chr', '') as Chromosome;
         }
-        if (!Object.keys(GET_CHROM_SIZES().interval).find(chr => chr === `chr${domain.chromosome}`)) {
+        if (!Object.keys(GET_CHROM_SIZES(assembly).interval).find(chr => chr === `chr${domain.chromosome}`)) {
             // we did not find any, so use '1' by default
             domain.chromosome = '1';
         }