Add Publication column for Cases, Biospecimen, and Files tabs

data/processSynapseJSON.log

@@ -2,9 +2,9 @@ ncc: Version 0.28.6
 ncc: Compiling file index.js
 ncc: Using [email protected] (local user-provided)
    40kB  sourcemap-register.js
-10201kB  index.js
- 1427kB  index.js.map
-10241kB  [26259ms] - ncc 0.28.6
+11086kB  index.js
+ 1428kB  index.js.map
+11126kB  [33746ms] - ncc 0.28.6
 Missing ParentBiospecimenID:  {
   Component: 'ScRNA-seqLevel3',
   Filename: 'single_cell_RNAseq_level_3_ped_glioma/HTAN_pHGG_161_New_Reg1_snRNA/barcodes.tsv.gz',

data/processSynapseJSON.ts

@@ -494,7 +494,7 @@ function processSynapseJSON(
         .mapValues((p) => new Set(getPublicationAssociatedParentDataFileIDs(p)))
         .value();
 
-    // add publication id
+    // add publication ids
     flatData.forEach((f) => {
         _.forEach(
             publicationParentDataFileIdsByUid,
@@ -612,6 +612,19 @@ function processSynapseJSON(
         ancestryByParticipantID
     );
 
+    // add publication ids for biospecimens
+    addBiopsecimenPublicationIds(files, filesById, biospecimenByBiospecimenID);
+
+    // add publication ids for participants
+    addParticipantPublicationIds(
+        biospecimenByBiospecimenID,
+        diagnosisByParticipantID
+    );
+    addParticipantPublicationIds(
+        biospecimenByBiospecimenID,
+        demographicsByParticipantID
+    );
+
     const dbgapImgSynapseSet = new Set<string>(
         getDbgapImgSynapseIds(entitiesById)
     );
@@ -733,6 +746,71 @@ function addPrimaryParents(
     });
 }
 
+function addParticipantPublicationIds(
+    biospecimenByBiospecimenID: {
+        [biospecimenID: string]: BaseSerializableEntity;
+    },
+    casesByParticipantID: {
+        [participantID: string]: BaseSerializableEntity;
+    }
+) {
+    // for each biospecimen find related participants and add publication ids
+    _.forEach(biospecimenByBiospecimenID, (biospecimen) => {
+        const participant = getParticipantCaseData(
+            biospecimen,
+            biospecimenByBiospecimenID,
+            casesByParticipantID
+        );
+
+        if (participant && !_.isEmpty(biospecimen.publicationIds)) {
+            participant.publicationIds = participant.publicationIds || [];
+            participant.publicationIds.push(...biospecimen.publicationIds!);
+        }
+    });
+
+    // remove duplicate publication ids if any
+    _.forEach(casesByParticipantID, (participant) => {
+        if (!_.isEmpty(participant.publicationIds)) {
+            participant.publicationIds = _.uniq(participant.publicationIds);
+        }
+    });
+}
+
+function addBiopsecimenPublicationIds(
+    files: BaseSerializableEntity[],
+    filesByHTANId: { [DataFileID: string]: BaseSerializableEntity },
+    biospecimenByBiospecimenID: {
+        [biospecimenID: string]: BaseSerializableEntity;
+    }
+) {
+    // for each file find related biospecimens and add publication ids
+    files.forEach((file) => {
+        const primaryParents =
+            file.primaryParents && file.primaryParents.length
+                ? file.primaryParents
+                : [file.DataFileID];
+        const biospecimens = getUniqueParentBiospecimens(
+            primaryParents,
+            filesByHTANId,
+            biospecimenByBiospecimenID
+        );
+
+        if (!_.isEmpty(file.publicationIds)) {
+            biospecimens.forEach((biospecimen) => {
+                biospecimen.publicationIds = biospecimen.publicationIds || [];
+                biospecimen.publicationIds.push(...file.publicationIds!);
+            });
+        }
+    });
+
+    // remove duplicate publication ids if any
+    _.forEach(biospecimenByBiospecimenID, (biospecimen) => {
+        if (!_.isEmpty(biospecimen.publicationIds)) {
+            biospecimen.publicationIds = _.uniq(biospecimen.publicationIds);
+        }
+    });
+}
+
 function findAndAddPrimaryParents(
     f: BaseSerializableEntity,
     filesByFileId: { [DataFileID: string]: BaseSerializableEntity }
@@ -865,6 +943,23 @@ function getParentBiospecimens(
     );
 }
 
+function getUniqueParentBiospecimens(
+    primaryParents: string[],
+    filesByHTANId: { [DataFileID: string]: BaseSerializableEntity },
+    biospecimenByBiospecimenID: {
+        [biospecimenID: string]: BaseSerializableEntity;
+    }
+) {
+    return _(primaryParents)
+        .map((p) =>
+            getParentBiospecimens(filesByHTANId[p], biospecimenByBiospecimenID)
+        )
+        .flatten()
+        .compact()
+        .uniqBy((b) => b.BiospecimenID)
+        .value();
+}
+
 function getSampleAndPatientData(
     file: BaseSerializableEntity,
     filesByHTANId: { [DataFileID: string]: BaseSerializableEntity },
@@ -899,14 +994,11 @@ function getSampleAndPatientData(
         }
     }
 
-    const biospecimen = _(primaryParents)
-        .map((p) =>
-            getParentBiospecimens(filesByHTANId[p], biospecimenByBiospecimenID)
-        )
-        .flatten()
-        .compact()
-        .uniqBy((b) => b.BiospecimenID)
-        .value();
+    const biospecimen = getUniqueParentBiospecimens(
+        primaryParents,
+        filesByHTANId,
+        biospecimenByBiospecimenID
+    );
 
     const diagnosis = _.uniqBy(
         getCaseData(
@@ -938,6 +1030,44 @@ function getSampleAndPatientData(
     return { biospecimen, diagnosis, demographics, therapy };
 }
 
+function getParticipantCaseData(
+    entity: BaseSerializableEntity,
+    biospecimenByBiospecimenID: {
+        [biospecimenID: string]: BaseSerializableEntity;
+    },
+    casesByParticipantID: {
+        [participantID: string]: BaseSerializableEntity;
+    }
+) {
+    // parentID can be both participant or biospecimen, so keep
+    // going up the tree until participant is found.
+    let parentID = entity.ParentID;
+    const alreadyProcessed = new Set();
+
+    while (parentID in biospecimenByBiospecimenID) {
+        // this is to prevent infinite loop due to possible circular references
+        if (alreadyProcessed.has(parentID)) {
+            break;
+        } else {
+            alreadyProcessed.add(parentID);
+        }
+
+        const parentBioSpecimen = biospecimenByBiospecimenID[parentID];
+        if (parentBioSpecimen.ParentID) {
+            parentID = parentBioSpecimen.ParentID;
+        }
+    }
+
+    if (!(parentID in casesByParticipantID)) {
+        // console.error(
+        //     `${s.BiospecimenID} does not have a parentID (${parentID}) with diagnosis/demographics information`
+        // );
+        return undefined;
+    } else {
+        return casesByParticipantID[parentID] as Entity;
+    }
+}
+
 function getCaseData(
     biospecimen: BaseSerializableEntity[],
     biospecimenByBiospecimenID: {
@@ -948,35 +1078,13 @@ function getCaseData(
     }
 ) {
     return biospecimen
-        .map((s) => {
-            // parentID can be both participant or biospecimen, so keep
-            // going up the tree until participant is found.
-            let parentID = s.ParentID;
-            const alreadyProcessed = new Set();
-
-            while (parentID in biospecimenByBiospecimenID) {
-                // this is to prevent infinite loop due to possible circular references
-                if (alreadyProcessed.has(parentID)) {
-                    break;
-                } else {
-                    alreadyProcessed.add(parentID);
-                }
-
-                const parentBioSpecimen = biospecimenByBiospecimenID[parentID];
-                if (parentBioSpecimen.ParentID) {
-                    parentID = parentBioSpecimen.ParentID;
-                }
-            }
-
-            if (!(parentID in casesByParticipantID)) {
-                // console.error(
-                //     `${s.BiospecimenID} does not have a parentID (${parentID}) with diagnosis/demographics information`
-                // );
-                return undefined;
-            } else {
-                return casesByParticipantID[parentID] as Entity;
-            }
-        })
+        .map((s) =>
+            getParticipantCaseData(
+                s,
+                biospecimenByBiospecimenID,
+                casesByParticipantID
+            )
+        )
         .filter((f) => !!f) as BaseSerializableEntity[];
 }
 

lib/helpers.ts

@@ -44,7 +44,7 @@ export async function fetchData(): Promise<LoadDataResult> {
     const processedSynURL =
         process.env.NODE_ENV === 'development'
             ? '/processed_syn_data.json'
-            : `${getCloudBaseUrl()}/processed_syn_data_20241226_1531.json`;
+            : `${getCloudBaseUrl()}/processed_syn_data_20250103_1351.json`;
     return fetchSynData(processedSynURL);
 }
 

packages/data-portal-commons/src/components/PublicationIcon.tsx

@@ -0,0 +1,35 @@
+import { FontAwesomeIcon } from '@fortawesome/react-fontawesome';
+import { faBook } from '@fortawesome/free-solid-svg-icons';
+import Tooltip from 'rc-tooltip';
+import React from 'react';
+
+import { PublicationManifest } from '../lib/entity';
+import {
+    getCiteFromPublicationManifest,
+    getPublicationUid,
+} from '../lib/publicationHelpers';
+
+export const PublicationIcon: React.FunctionComponent<{
+    publicationManifest: PublicationManifest;
+}> = (props) => {
+    const { publicationManifest } = props;
+
+    return (
+        <Tooltip
+            overlay={getCiteFromPublicationManifest(publicationManifest)}
+            key={getPublicationUid(publicationManifest)}
+        >
+            <a
+                href={`//${
+                    window.location.host
+                }/publications/${getPublicationUid(publicationManifest)}`}
+                key={getPublicationUid(publicationManifest)}
+                style={{ paddingRight: 3 }}
+            >
+                <FontAwesomeIcon icon={faBook} />
+            </a>
+        </Tooltip>
+    );
+};
+
+export default PublicationIcon;

packages/data-portal-commons/src/index.ts

@@ -12,6 +12,7 @@ export * from './lib/types';
 
 export * from './components/AtlasDescription';
 export * from './components/ExpandableText';
+export * from './components/PublicationIcon';
 export * from './components/ViewDetailsModal';
 
 export { default as commonStyles } from './components/common.module.scss';

packages/data-portal-explore/src/components/AtlasTable.tsx

@@ -18,9 +18,8 @@ import {
     AtlasDescription,
     AtlasMetaData,
     Entity,
-    getCiteFromPublicationManifest,
-    getPublicationUid,
     isManuscriptInReview,
+    PublicationIcon,
     PublicationManifest,
 } from '@htan/data-portal-commons';
 import { ExploreTab } from '../lib/types';
@@ -270,26 +269,9 @@ export class AtlasTable extends React.Component<IAtlasTableProps> {
                     if (atlasTableData.publicationManifests.length > 0) {
                         return atlasTableData.publicationManifests.map(
                             (publicationManifest) => (
-                                <Tooltip
-                                    overlay={getCiteFromPublicationManifest(
-                                        publicationManifest
-                                    )}
-                                    key={getPublicationUid(publicationManifest)}
-                                >
-                                    <a
-                                        href={`//${
-                                            window.location.host
-                                        }/publications/${getPublicationUid(
-                                            publicationManifest
-                                        )}`}
-                                        key={getPublicationUid(
-                                            publicationManifest
-                                        )}
-                                        style={{ paddingRight: 3 }}
-                                    >
-                                        <FontAwesomeIcon icon={faBook} />
-                                    </a>
-                                </Tooltip>
+                                <PublicationIcon
+                                    publicationManifest={publicationManifest}
+                                />
                             )
                         );
                     } else {

packages/data-portal-explore/src/components/BiospecimenTable.tsx

@@ -5,13 +5,14 @@ import {
     EnhancedDataTable,
     getDefaultDataTableStyle,
 } from '@htan/data-portal-table';
-import { Atlas, Entity } from '@htan/data-portal-commons';
+import { Atlas, Entity, PublicationManifest } from '@htan/data-portal-commons';
 import { GenericAttributeNames } from '@htan/data-portal-utils';
 import { DataSchemaData, SchemaDataId } from '@htan/data-portal-schema';
 
 import {
     generateColumnsForDataSchema,
     getAtlasColumn,
+    getPublicationColumn,
     sortByBiospecimenId,
     sortByParentID,
 } from '../lib/dataTableHelpers';
@@ -21,6 +22,7 @@ interface IBiospecimenTableProps {
     synapseAtlases: Atlas[];
     schemaDataById?: { [schemaDataId: string]: DataSchemaData };
     genericAttributeMap?: { [attr: string]: GenericAttributeNames };
+    publicationsByUid?: { [uid: string]: PublicationManifest };
 }
 
 export const BiospecimenTable: React.FunctionComponent<IBiospecimenTableProps> = (
@@ -46,11 +48,12 @@ export const BiospecimenTable: React.FunctionComponent<IBiospecimenTableProps> =
         columns,
         (c) => c.selector === GenericAttributeNames.BiospecimenID
     );
-    // insert Atlas Name right after Biospecimen ID
+    // insert Atlas Name and Publications right after Biospecimen ID
     columns.splice(
         indexOfBiospecimenId + 1,
         0,
-        getAtlasColumn(props.synapseAtlases)
+        getAtlasColumn(props.synapseAtlases),
+        getPublicationColumn(props.publicationsByUid)
     );
 
     return (