DRKG predicate mappings

[schatzkara]

As part of a research project, I am using DRKG (The Drug Repurposing Knowledge Graph) and am hoping to map it to biolink so that I can do some analysis comparing DRKG and ROBOKOP.

Would it be possible to add predicate mappings for the relations in DRKG that are not already mapped? They are listed below.

• DRUGBANK::carrier::Compound:Gene
• DRUGBANK::ddi-interactor-in::Compound:Compound
• DRUGBANK::enzyme::Compound:Gene
• DRUGBANK::target::Compound:Gene
• DRUGBANK::x-atc::Compound:Atc
• DRUGBANK::treats::Compound:Disease
• GNBR::A+::Compound:Gene
• GNBR::A-::Compound:Gene
• GNBR::B::Compound:Gene
• GNBR::B::Gene:Gene
• GNBR::C::Compound:Disease
• GNBR::D::Gene:Disease
• GNBR::E+::Compound:Gene
• GNBR::E+::Gene:Gene
• GNBR::E-::Compound:Gene
• GNBR::E::Compound:Gene
• GNBR::E::Gene:Gene
• GNBR::G::Gene:Disease
• GNBR::H::Gene:Gene
• GNBR::I::Gene:Gene
• GNBR::J::Compound:Disease
• GNBR::J::Gene:Disease
• GNBR::K::Compound:Gene
• GNBR::L::Gene:Disease
• GNBR::Md::Gene:Disease
• GNBR::Mp::Compound:Disease
• GNBR::N::Compound:Gene
• GNBR::O::Compound:Gene
• GNBR::Pa::Compound:Disease
• GNBR::Pr::Compound:Disease
• GNBR::Q::Gene:Gene
• GNBR::Rg::Gene:Gene
• GNBR::Sa::Compound:Disease
• GNBR::T::Compound:Disease
• GNBR::Te::Gene:Disease
• GNBR::U::Gene:Disease
• GNBR::Ud::Gene:Disease
• GNBR::V+::Gene:Gene
• GNBR::W::Gene:Gene
• GNBR::X::Gene:Disease
• GNBR::Y::Gene:Disease
• GNBR::Z::Compound:Gene
• GNBR::in_tax::Gene:Tax
• Hetionet::AdG::Anatomy:Gene
• Hetionet::AeG::Anatomy:Gene
• Hetionet::AuG::Anatomy:Gene
• Hetionet::CbG::Compound:Gene
• Hetionet::CcSE::Compound:Side Effect
• Hetionet::CdG::Compound:Gene
• Hetionet::CpD::Compound:Disease
• Hetionet::CrC::Compound:Compound
• Hetionet::CtD::Compound:Disease
• Hetionet::CuG::Compound:Gene
• Hetionet::DaG::Disease:Gene
• Hetionet::DdG::Disease:Gene
• Hetionet::DlA::Disease:Anatomy
• Hetionet::DpS::Disease:Symptom
• Hetionet::DrD::Disease:Disease
• Hetionet::DuG::Disease:Gene
• Hetionet::GcG::Gene:Gene
• Hetionet::GiG::Gene:Gene
• Hetionet::GpBP::Gene:Biological Process
• Hetionet::GpCC::Gene:Cellular Component
• Hetionet::GpMF::Gene:Molecular Function
• Hetionet::GpPW::Gene:Pathway
• Hetionet::Gr>G::Gene:Gene
• Hetionet::PCiC::Pharmacologic Class:Compound
• INTACT::ADP RIBOSYLATION REACTION::Gene:Gene
• INTACT::ASSOCIATION::Compound:Gene
• INTACT::ASSOCIATION::Gene:Gene
• INTACT::CLEAVAGE REACTION::Gene:Gene
• INTACT::COLOCALIZATION::Gene:Gene
• INTACT::DEPHOSPHORYLATION REACTION::Gene:Gene
• INTACT::DIRECT INTERACTION::Compound:Gene
• INTACT::DIRECT INTERACTION::Gene:Gene
• INTACT::PHOSPHORYLATION REACTION::Gene:Gene
• INTACT::PHYSICAL ASSOCIATION::Compound:Gene
• INTACT::PHYSICAL ASSOCIATION::Gene:Gene
• INTACT::PROTEIN CLEAVAGE::Gene:Gene
• INTACT::UBIQUITINATION REACTION::Gene:Gene
• STRING::ACTIVATION::Gene:Gene
• STRING::BINDING::Gene:Gene
• STRING::CATALYSIS::Gene:Gene
• STRING::EXPRESSION::Gene:Gene
• STRING::INHIBITION::Gene:Gene
• STRING::OTHER::Gene:Gene
• STRING::PTMOD::Gene:Gene
• STRING::REACTION::Gene:Gene
• bioarx::Coronavirus_ass_host_gene::Disease:Gene
• bioarx::Covid2_acc_host_gene::Disease:Gene
• bioarx::DrugHumGen:Compound:Gene
• bioarx::DrugVirGen:Compound:Gene
• bioarx::HumGenHumGen:Gene:Gene
• bioarx::VirGenHumGen:Gene:Gene

@balhoff

[cmungall]

Can you help us understand the format?

Are these lists of actual predicates

I am not sure how to intrepret:

GNBR::E+::Compound:Gene GNBR::E+::Gene:Gene GNBR::E-::Compound:Gene GNBR::E::Compound:Gene

[schatzkara]

Yes, of course. Sorry I should have added this to the initial issue. The format is: source::predicate::connected_entity_type1:connected_entity_type2

Specifically, the ones that you mentioned come from GNBR which uses the following key for the predicate types. GNBR Predicate Key

The predicates from Hetionet are similarly confusing and are defined by the following key: Hetionet Predicate Key

DRKG also provides some (limited) descriptions and interaction types for the predicates, which is in the attached file. relation_glossary.zip

[sierra-moxon]

@schatzkara - I imagine we'd use full predicate names like 'agonism' vs. 'A+' (we could potentially add things like 'A+' as an alias? - would this work?)

[schatzkara]

Yes, full names would definitely be fine as opposed to the far less readable acronyms :)

[mbrush]

Note that some of these predicates fall into the 'interaction mechanism' category, which the Chemical WG is working to harmonize. so work on these should be seeded to / coordinated with them.

Their working s/s is HERE. Note that GNBR is not one of the sources they targeted for harmonization - so we should be sure to add the GNBR interaction mechanism terms to the harmonization s/s . I added a stub sheet to the doc with a link to the GNBR predicate list.

[sierra-moxon]

Thanks @mbrush! Work added on Matt's spreadsheet: https://docs.google.com/spreadsheets/d/1VI2s5YVDwOHiQv7hBWkl12d86yWvm-z_pDJFnDwsoiY/edit#gid=542956250

[sierra-moxon]

tagging @ericawood as she had a ticket that was very similar here: #679

[nlharris]

Can this be closed?

[sierra-moxon]

@schatzkar, I believe much of this case is handled by this PR (refactoring the chemical hierarchy). Would your team have a chance to review that PR with regard to this ticket? I've attached this ticket there, so that when the PR is merged, this ticket will be resolved.

[schatzkara]

@schatzkar, I believe much of this case is handled by this PR (refactoring the chemical hierarchy). Would your team have a chance to review that PR with regard to this ticket? I've attached this ticket there, so that when the PR is merged, this ticket will be resolved.

@sierra-moxon It looks like some of the DRUGBANK predicates and several of the Hetionet predicates are covered there, but some others are not. I did not see the GNBR, INTACT, STRING, or bioarx predicates there. I can give a more specific list if that's what you're looking for. Maybe the rest do not fit into the ontology?