entity-source coloring (replaces gene coloring)

data/mmcif-field-names.csv

@@ -93,12 +93,24 @@ entity_poly_seq.hetero
 
 entity_src_gen.entity_id
 entity_src_gen.pdbx_src_id
-entity_src_gen.pdbx_alt_source_flag 
-entity_src_gen.pdbx_seq_type 
+entity_src_gen.pdbx_alt_source_flag
+entity_src_gen.pdbx_seq_type
 entity_src_gen.pdbx_beg_seq_num
 entity_src_gen.pdbx_end_seq_num
 entity_src_gen.pdbx_gene_src_gene
 
+entity_src_nat.entity_id
+entity_src_nat.pdbx_src_id
+entity_src_nat.pdbx_alt_source_flag
+entity_src_nat.pdbx_beg_seq_num
+entity_src_nat.pdbx_end_seq_num
+
+pdbx_entity_src_syn.entity_id
+pdbx_entity_src_syn.pdbx_src_id
+pdbx_entity_src_syn.pdbx_alt_source_flag
+pdbx_entity_src_syn.pdbx_beg_seq_num
+pdbx_entity_src_syn.pdbx_end_seq_num
+
 pdbx_entity_branch.entity_id
 pdbx_entity_branch.type
 

src/mol-io/reader/cif/schema/mmcif.ts

@@ -1,7 +1,7 @@
 /**
  * Copyright (c) 2017-2018 mol* contributors, licensed under MIT, See LICENSE file for more info.
  *
- * Code-generated 'mmCIF' schema file. Dictionary versions: mmCIF 5.309, IHM 0.141, CARB draft.
+ * Code-generated 'mmCIF' schema file. Dictionary versions: mmCIF 5.310, IHM 0.141, CARB draft.
  *
  * @author mol-star package (src/apps/schema-generator/generate)
  */
@@ -1964,6 +1964,40 @@ export const mmCIF_Schema = {
          */
         details: str,
     },
+    /**
+     * Data items in the ENTITY_SRC_NAT category record details of
+     * the source from which the entity was obtained in cases
+     * where the entity was isolated directly from a natural tissue.
+     */
+    entity_src_nat: {
+        /**
+         * This data item is a pointer to _entity.id in the ENTITY category.
+         */
+        entity_id: str,
+        /**
+         * This data item is an ordinal identifier for entity_src_nat data records.
+         */
+        pdbx_src_id: int,
+        /**
+         * This data item identifies cases in which an alternative source
+         * modeled.
+         */
+        pdbx_alt_source_flag: Aliased<'sample' | 'model'>(str),
+        /**
+         * The beginning polymer sequence position for the polymer section corresponding
+         * to this source.
+         *
+         * A reference to the sequence position in the entity_poly category.
+         */
+        pdbx_beg_seq_num: int,
+        /**
+         * The ending polymer sequence position for the polymer section corresponding
+         * to this source.
+         *
+         * A reference to the sequence position in the entity_poly category.
+         */
+        pdbx_end_seq_num: int,
+    },
     /**
      * Data items in the ENTITY_SRC_GEN category record details of
      * the source from which the entity was obtained in cases
@@ -2010,6 +2044,39 @@ export const mmCIF_Schema = {
          */
         pdbx_end_seq_num: int,
     },
+    /**
+     * The data items in category PDBX_ENTITY_SRC_SYN record the source details
+     * about chemically synthesized molecules.
+     */
+    pdbx_entity_src_syn: {
+        /**
+         * This data item is a pointer to _entity.id in the ENTITY category.
+         */
+        entity_id: str,
+        /**
+         * This data item is an ordinal identifier for pdbx_entity_src_syn data records.
+         */
+        pdbx_src_id: int,
+        /**
+         * This data item identifies cases in which an alternative source
+         * modeled.
+         */
+        pdbx_alt_source_flag: Aliased<'sample' | 'model'>(str),
+        /**
+         * The beginning polymer sequence position for the polymer section corresponding
+         * to this source.
+         *
+         * A reference to the sequence position in the entity_poly category.
+         */
+        pdbx_beg_seq_num: int,
+        /**
+         * The ending polymer sequence position for the polymer section corresponding
+         * to this source.
+         *
+         * A reference to the sequence position in the entity_poly category.
+         */
+        pdbx_end_seq_num: int,
+    },
     /**
      * Data items in the PDBX_ENTITY_DESCRIPTOR category provide
      * string descriptors of entity chemical structure.

src/mol-theme/color.ts

@@ -27,7 +27,7 @@ import { UnitIndexColorThemeProvider } from './color/unit-index';
 import { ScaleLegend } from 'mol-util/color/scale';
 import { TableLegend } from 'mol-util/color/tables';
 import { UncertaintyColorThemeProvider } from './color/uncertainty';
-import { GeneColorThemeProvider } from './color/gene';
+import { EntitySourceColorThemeProvider } from './color/entity-source';
 import { IllustrativeColorThemeProvider } from './color/illustrative';
 import { HydrophobicityColorThemeProvider } from './color/hydrophobicity';
 
@@ -76,7 +76,7 @@ export const BuiltInColorThemes = {
     'cross-link': CrossLinkColorThemeProvider,
     'element-index': ElementIndexColorThemeProvider,
     'element-symbol': ElementSymbolColorThemeProvider,
-    'gene': GeneColorThemeProvider,
+    'entity-source': EntitySourceColorThemeProvider,
     'hydrophobicity': HydrophobicityColorThemeProvider,
     'illustrative': IllustrativeColorThemeProvider,
     'molecule-type': MoleculeTypeColorThemeProvider,

src/mol-theme/color/entity-source.ts

@@ -0,0 +1,143 @@
+/**
+ * Copyright (c) 2019 mol* contributors, licensed under MIT, See LICENSE file for more info.
+ *
+ * @author Alexander Rose <alexander.rose@weirdbyte.de>
+ */
+
+import { StructureProperties, StructureElement, Link, Model } from 'mol-model/structure';
+import { ColorScale, Color } from 'mol-util/color';
+import { Location } from 'mol-model/location';
+import { ColorTheme, LocationColor } from '../color';
+import { ParamDefinition as PD } from 'mol-util/param-definition'
+import { ThemeDataContext } from 'mol-theme/theme';
+import { ColorListOptions, ColorListName } from 'mol-util/color/scale';
+import { Table, Column } from 'mol-data/db';
+import { mmCIF_Schema } from 'mol-io/reader/cif/schema/mmcif';
+
+const DefaultColor = Color(0xCCCCCC)
+const Description = 'Gives ranges of a polymer chain a color based on the entity source it originates from. Genes get the same color per entity'
+
+export const EntitySourceColorThemeParams = {
+    list: PD.ColorScale<ColorListName>('RedYellowBlue', ColorListOptions),
+}
+export type EntitySourceColorThemeParams = typeof EntitySourceColorThemeParams
+export function getEntitySourceColorThemeParams(ctx: ThemeDataContext) {
+    return EntitySourceColorThemeParams // TODO return copy
+}
+
+function modelEntityKey(modelIndex: number, entityId: string) {
+    return `${modelIndex}|${entityId}`
+}
+
+type EntitySrc = Table<{
+    entity_id: mmCIF_Schema['entity_src_gen']['entity_id'],
+    pdbx_src_id: mmCIF_Schema['entity_src_gen']['pdbx_src_id'],
+    pdbx_beg_seq_num: mmCIF_Schema['entity_src_gen']['pdbx_beg_seq_num'],
+    pdbx_end_seq_num: mmCIF_Schema['entity_src_gen']['pdbx_end_seq_num'],
+}>
+type GeneSrcGene = Column<mmCIF_Schema['entity_src_gen']['pdbx_gene_src_gene']['T']>
+
+function srcKey(modelIndex: number, entityId: string, srcId: number, gene: string) {
+    return `${modelIndex}|${entityId}|${gene ? gene : srcId}`
+}
+
+function addSrc(seqToSrcByModelEntity: Map<string, Int16Array>, srcKeySerialMap: Map<string, number>, modelIndex: number, model: Model, entity_src: EntitySrc, gene_src_gene?: GeneSrcGene) {
+    const { entity_id, pdbx_src_id, pdbx_beg_seq_num, pdbx_end_seq_num } = entity_src
+    for (let j = 0, jl = entity_src._rowCount; j < jl; ++j) {
+        const entityId = entity_id.value(j)
+        const mK = modelEntityKey(modelIndex, entityId)
+        let seqToSrc: Int16Array
+        if (!seqToSrcByModelEntity.has(mK)) {
+            const entityIndex = model.entities.getEntityIndex(entityId)
+            const seq = model.sequence.sequences[entityIndex].sequence
+            seqToSrc = new Int16Array(seq.sequence.length)
+            seqToSrcByModelEntity.set(mK, seqToSrc)
+        } else {
+            seqToSrc = seqToSrcByModelEntity.get(mK)!
+        }
+        const sK = srcKey(modelIndex, entityId, pdbx_src_id.value(j), gene_src_gene ? gene_src_gene.value(j).join(',') : '')
+
+        // may not be given (= 0) indicating src is for the whole seq
+        const beg = pdbx_beg_seq_num.valueKind(j) === Column.ValueKind.Present ? pdbx_beg_seq_num.value(j) : 1
+        const end = pdbx_end_seq_num.valueKind(j) === Column.ValueKind.Present ? pdbx_end_seq_num.value(j) : seqToSrc.length
+
+        let srcIndex: number // serial no starting from 1
+        if (srcKeySerialMap.has(sK)) {
+            srcIndex = srcKeySerialMap.get(sK)!
+        } else {
+            srcIndex = srcKeySerialMap.size + 1
+            srcKeySerialMap.set(sK, srcIndex)
+        }
+        // set src index
+        for (let i = beg, il = end; i <= il; ++i) {
+            seqToSrc[i - 1] = srcIndex
+        }
+    }
+}
+
+export function EntitySourceColorTheme(ctx: ThemeDataContext, props: PD.Values<EntitySourceColorThemeParams>): ColorTheme<EntitySourceColorThemeParams> {
+    let color: LocationColor
+    const scale = ColorScale.create({ listOrName: props.list, minLabel: 'Start', maxLabel: 'End' })
+    const { structure } = ctx
+
+    if (structure) {
+        const l = StructureElement.create()
+        const { models } = structure
+        const seqToSrcByModelEntity = new Map<string, Int16Array>()
+        const srcKeySerialMap = new Map<string, number>() // serial no starting from 1
+
+        for (let i = 0, il = models.length; i <il; ++i) {
+            const m = models[i]
+            if (m.sourceData.kind !== 'mmCIF') continue
+            const { entity_src_gen, entity_src_nat, pdbx_entity_src_syn } = m.sourceData.data
+            addSrc(seqToSrcByModelEntity, srcKeySerialMap, i, m, entity_src_gen, entity_src_gen.pdbx_gene_src_gene)
+            addSrc(seqToSrcByModelEntity, srcKeySerialMap, i, m, entity_src_nat)
+            addSrc(seqToSrcByModelEntity, srcKeySerialMap, i, m, pdbx_entity_src_syn)
+        }
+        scale.setDomain(1, srcKeySerialMap.size)
+        const scaleColor = scale.color
+
+        const getSrcColor = (location: StructureElement) => {
+            const modelIndex = structure.models.indexOf(location.unit.model)
+            const entityId = StructureProperties.entity.id(location)
+            const mK = modelEntityKey(modelIndex, entityId)
+            const seqToSrc = seqToSrcByModelEntity.get(mK)
+            if (seqToSrc) {
+                // minus 1 to convert seqId to array index
+                return scaleColor(seqToSrc[StructureProperties.residue.label_seq_id(location) - 1])
+            } else {
+                return DefaultColor
+            }
+        }
+
+        color = (location: Location): Color => {
+            if (StructureElement.isLocation(location)) {
+                return getSrcColor(location)
+            } else if (Link.isLocation(location)) {
+                l.unit = location.aUnit
+                l.element = location.aUnit.elements[location.aIndex]
+                return getSrcColor(l)
+            }
+            return DefaultColor
+        }
+    } else {
+        color = () => DefaultColor
+    }
+
+    return {
+        factory: EntitySourceColorTheme,
+        granularity: 'group',
+        color,
+        props,
+        description: Description,
+        legend: scale ? scale.legend : undefined
+    }
+}
+
+export const EntitySourceColorThemeProvider: ColorTheme.Provider<EntitySourceColorThemeParams> = {
+    label: 'Entity Source',
+    factory: EntitySourceColorTheme,
+    getParams: getEntitySourceColorThemeParams,
+    defaultValues: PD.getDefaultValues(EntitySourceColorThemeParams),
+    isApplicable: (ctx: ThemeDataContext) => !!ctx.structure
+}

src/mol-theme/color/gene.ts

@@ -1,127 +0,0 @@
-/**
- * Copyright (c) 2019 mol* contributors, licensed under MIT, See LICENSE file for more info.
- *
- * @author Alexander Rose <alexander.rose@weirdbyte.de>
- */
-
-import { StructureProperties, StructureElement, Link } from 'mol-model/structure';
-import { ColorScale, Color } from 'mol-util/color';
-import { Location } from 'mol-model/location';
-import { ColorTheme, LocationColor } from '../color';
-import { ParamDefinition as PD } from 'mol-util/param-definition'
-import { ThemeDataContext } from 'mol-theme/theme';
-import { ColorListOptions, ColorListName } from 'mol-util/color/scale';
-import { NumberArray } from 'mol-util/type-helpers';
-
-const DefaultColor = Color(0xCCCCCC)
-const Description = 'Gives ranges of a polymer chain a color based on the gene (or linker/terminal extension) it originates from.'
-
-export const GeneColorThemeParams = {
-    list: PD.ColorScale<ColorListName>('RedYellowBlue', ColorListOptions),
-}
-export type GeneColorThemeParams = typeof GeneColorThemeParams
-export function getGeneColorThemeParams(ctx: ThemeDataContext) {
-    return GeneColorThemeParams // TODO return copy
-}
-
-function modelEntityKey(modelIndex: number, entityId: string) {
-    return `${modelIndex}|${entityId}`
-}
-
-function addGene(geneSerialMap: Map<string, number>, geneNames: string[], beg: number, end: number, seqToSrcGen: NumberArray) {
-    const gene = geneNames.map(s => s.toUpperCase()).sort().join(',')
-    let geneIndex = 0 // serial no starting from 1
-    if (gene === '') {
-        geneIndex = geneSerialMap.size + 1
-        geneSerialMap.set(`UNKNOWN${geneIndex}`, geneIndex)
-    } else if (geneSerialMap.has(gene)) {
-        geneIndex = geneSerialMap.get(gene)!
-    } else {
-        geneIndex = geneSerialMap.size + 1
-        geneSerialMap.set(gene, geneIndex)
-    }
-    for (let i = beg, il = end; i <= il; ++i) {
-        seqToSrcGen[i - 1] = geneIndex
-    }
-}
-
-export function GeneColorTheme(ctx: ThemeDataContext, props: PD.Values<GeneColorThemeParams>): ColorTheme<GeneColorThemeParams> {
-    let color: LocationColor
-    const scale = ColorScale.create({ listOrName: props.list, minLabel: 'Start', maxLabel: 'End' })
-    const { structure } = ctx
-
-    if (structure) {
-        const l = StructureElement.create()
-        const { models } = structure
-        const seqToSrcGenByModelEntity = new Map<string, NumberArray>()
-        const geneSerialMap = new Map<string, number>() // serial no starting from 1
-        for (let i = 0, il = models.length; i <il; ++i) {
-            const m = models[i]
-            if (m.sourceData.kind !== 'mmCIF') continue
-            const { entity_src_gen } = m.sourceData.data
-            
-            const { entity_id, pdbx_beg_seq_num, pdbx_end_seq_num, pdbx_gene_src_gene } = entity_src_gen
-            for (let j = 0, jl = entity_src_gen._rowCount; j < jl; ++j) {
-                const entityId = entity_id.value(j)
-                const k = modelEntityKey(i, entityId)
-                if (!seqToSrcGenByModelEntity.has(k)) {
-                    const entityIndex = m.entities.getEntityIndex(entityId)
-                    const seq = m.sequence.sequences[entityIndex].sequence
-                    const seqLength = seq.sequence.length
-                    const seqToGene = new Int16Array(seqLength)
-                    addGene(geneSerialMap, pdbx_gene_src_gene.value(j), pdbx_beg_seq_num.value(j), pdbx_end_seq_num.value(j), seqToGene)
-                    seqToSrcGenByModelEntity.set(k, seqToGene)
-                } else {
-                    const seqToGene = seqToSrcGenByModelEntity.get(k)!
-                    addGene(geneSerialMap, pdbx_gene_src_gene.value(j), pdbx_beg_seq_num.value(j), pdbx_end_seq_num.value(j), seqToGene)
-                    seqToSrcGenByModelEntity.set(k, seqToGene)
-                }
-            }
-        }
-        scale.setDomain(1, geneSerialMap.size)
-        const scaleColor = scale.color
-
-        const getGeneColor = (location: StructureElement) => {
-            const modelIndex = structure.models.indexOf(location.unit.model)
-            const entityId = StructureProperties.entity.id(location)
-            const k = modelEntityKey(modelIndex, entityId)
-            const seqToGene = seqToSrcGenByModelEntity.get(k)
-            if (seqToGene) {
-                // minus 1 to convert seqId to array index
-                return scaleColor(seqToGene[StructureProperties.residue.label_seq_id(location) - 1])
-            } else {
-                return DefaultColor
-            }
-        }
-
-        color = (location: Location): Color => {
-            if (StructureElement.isLocation(location)) {
-                return getGeneColor(location)
-            } else if (Link.isLocation(location)) {
-                l.unit = location.aUnit
-                l.element = location.aUnit.elements[location.aIndex]
-                return getGeneColor(l)
-            }
-            return DefaultColor
-        }
-    } else {
-        color = () => DefaultColor
-    }
-
-    return {
-        factory: GeneColorTheme,
-        granularity: 'group',
-        color,
-        props,
-        description: Description,
-        legend: scale ? scale.legend : undefined
-    }
-}
-
-export const GeneColorThemeProvider: ColorTheme.Provider<GeneColorThemeParams> = {
-    label: 'Gene',
-    factory: GeneColorTheme,
-    getParams: getGeneColorThemeParams,
-    defaultValues: PD.getDefaultValues(GeneColorThemeParams),
-    isApplicable: (ctx: ThemeDataContext) => !!ctx.structure
-}