Changes in EpiGraphDB v1.0

Here we document the changes regarding the EpiGraphDB platform and database from version 0.3 (2020-04) to version 1.0 (2021-03).

Overall architecture and components

• There is a major overhaul in the data integration pipeline and is now open sourced as epigraphdb-graph.

---

Schema and entity representation

General changes

• meta properties: each meta entity contains the following meta properties,
  • _id: unique id
  • _name: human readable label
  • _source: data source(s) of the entity

Meta nodes

• Meta: We have dropped the use of a Meta node for metadata information.

• Disease

• Drug

• Efo

• Event: Removed due to overhaul of source Reactome

• Gene
  • new props: biomart_source

• Gwas
  • new props: build

• Literature
  • new props: doi
  • renamed props:
    • pubmid_id: id

• Pathway
  • removed props: in_disease
  • renamed props:
    • reactome_id: id

• Protein
  • new props: name

• SemmedTriple: Renamed to LiteratureTriple. Entity labels are now removed from LiteratureTriples and only CUI id remain. Users will need to match from LiteratureTerm to get the entity labels.
  • removed props: count, subject_name, object_name, subject_type, object_type
  • new props: name (merge subject_name, predicate, object_name together)

• SemmedTerm: Renamed to LiteratureTerm
  • removed props: count

• Tissue
  • renamed props: tissue: {id, name}

• Variant
  • new props: pos, chr, ref, alt, build

Meta rels

• BN_GEN_COR: Renamed to GEN_COR
  • renamed props:
    • se: rg_SE
    • z: Z
    • h2_int: h2_intercept
    • h2_int_se: h2_intercept_SE
  • new props: rpheno, rg_intercept, rg_intercept_SE
  • removed props
    • h2_obs, h2_obs_se
    • gcov_int, gcov_int_se

• CPIC

• EFO_CHILD_OF

• EVENT_IN_PATHWAY: Removed due to overhauled source Reactome

• EXPRESSED_IN

• GENE_TO_LITERATURE: Removed due to overhauled source Reactome

• GENE_TO_PROTEIN

• GWAS_NLP

• GWAS_NLP_EFO

• GWAS_SEM: Renamed to GWAS_TO_LITERATURE_TRIPLE

• GWAS_TO_LIT: Renamed to GWAS_TO_LITERATURE

• GWAS_TO_VARIANT

• INTACT_INTERACTS_WITH_GENE_GENE: Removed due to obsolete source IntACT

• INTACT_INTERACTS_WITH_GENE_PROTEIN: Removed due to obsolete source IntACT

• INTACT_INTERACTS_WITH_PROTEIN_PROTEIN: Removed due to obsolete source IntACT

• INTACT_NOT_INTERACTS_WITH: Removed due to obsolete source IntACT

• METAMAP_LITE
  • new props: mesh

• MONDO_MAP_EFO

• MONDO_MAP_UMLS

• MR: Renamed to MR_EVE_MR (which only consists of precomputed results from source MR EvE)
  • removed props: log10pval

• OBS_COR

• OPENTARGETS_DRUG_TO_DISEASE

• OPENTARGETS_DRUG_TO_TARGET

• PATHWAY_TO_DISEASE: Removed due to overhauled source Reactome

• PATHWAY_TO_LITERATURE: Removed due to overhauled source Reactome

• PRECEDING_EVENT: Removed due to overhauled source Reactome

• PROTEIN_IN_EVENT: Removed due to overhauled source Reactome

• PROTEIN_TO_DISEASE: Removed due to overhauled source Reactome

• PROTEIN_TO_LITERATURE: Removed due to overhauled source Reactome

• PROTEIN_IN_PATHWAY

• PRS

• SEM_GENE: Renamed to TERM_TO_GENE

• SEM_OBJ: Renamed to SEMMEDDB_OBJ

• SEM_PREDICATE: Renamed to SEMMEDDB_PREDICATE

• SEM_SUB: Renamed to SEMMEDDB_SUB

• SEM_TO_LIT: Renamed to SEMMEDDB_TO_LIT

• STRING_INTERACT_WITH

• TOPHITS: Renamed to OPENGWAS_TOPHITS

• VARIANT_TO_GENE

• XQTL_MULTI_SNP_MR

• XQTL_SINGLE_SNP_MR_GENE_GWAS

• XQTL_SINGLE_SNP_MR_SNP_GENE

New meta rels

• BIORXIV_OBJ, BIORXIV_PREDICATE, BIORXIV_SUB, BIORXIV_TO_LIT
• MEDRXIV_OBJ, MEDRXIV_PREDICATE, MEDRXIV_SUB, MEDRXIV_TO_LIT
• GWAS_EFO_EBI
• PATHWAY_CHILD_OF

---

Data sources

New data sources

• Annotation of EBI GWAS traits to EFO
• ClinVar

Removed data sources

• IntAct: We removed IntAct as its import pipeline to EpiGraphDB is not being maintained at the moment. We will keep this under review.

Overhauled / updated data sources

• Reactome
• SemMedDB
• MR-EvE

---

Graph database and cypher

• EpiGraphDB graph database is now built on neo4j v4, as a side effect the following cypher will no longer return parsible dictionary structure to the API MATCH (n)-[r]-(m) RETURN n, r, m, instead use MATCH (n)-[r]-(m) RETURN n, properties(r) as r, m

---

API

Changes in the API are generally compatibility updates regarding the database.

Endpoint changes:

• POST /cypher/builder/plain: Removed. Users can use POST /cypher to send cypher queries.

---

Web UI

We have added a search page to display results from the header search bar.

---

R package

Package version 0.2.2 now works with EpiGraphDB platform 1.0.0.