Add Drug Mechanism of Action information from ChEMBL

[danidi]

In ChEMBL, the connection of a drug compound to it's target is available via the drug_mechanism field. Currently, the use-case identifying targets for known drugs is not possible with the system. Either adding the drug mechanism information (e.g. to Compound Information), or the drugbank information (see https://github.com/openphacts/GLOBAL/issues/233), or both, would really enhance the system.

[ChristineChichester]

The mechanism of action would be a nice addition. Edgar also mentioned this in Santiago.

Although the use case for identifying the known drugs that are associated with targets can be solved (drug_type filter). It is available in Target Info from DrugBank ?db_compound drugbank:genericName ?drug_name ; drugbank:drugType ?drugType ;

On Feb 20, 2015, at 4:26 PM, danidi notifications@github.com wrote:

In ChEMBL, the connection of a drug compound to it's target is available via the drug_mechanism field. Currently, the use-case identifying targets for known drugs is not possible with the system. Either adding the drug mechanism information (e.g. to Compound Information), or the drugbank information (see #233 https://github.com/openphacts/GLOBAL/issues/233), or both, would really enhance the system.

— Reply to this email directly or view it on GitHub https://github.com/openphacts/GLOBAL/issues/235.

[Chris-Evelo]

Yes, absolutely.

But I suppose we will get the information automatically when we update to the latest ChEMBL and Drugbank versions. The question would be how to harmonise that information across the two resources and how we expose that again on the API. Any ideas on that?

 Best, Chris

— Prof.Dr. Chris Evelo Dept. Head Bioinformatics – BiGCaT Maastricht University

From: ChristineChichester notifications@github.com<mailto:notifications@github.com> Reply-To: openphacts/GLOBAL reply@reply.github.com<mailto:reply@reply.github.com> Date: Sunday 22 February 2015 8:13 To: openphacts/GLOBAL GLOBAL@noreply.github.com<mailto:GLOBAL@noreply.github.com> Subject: Re: [GLOBAL] Add Drug Mechanism of Action information from ChEMBL (#235)

The mechanism of action would be a nice addition. Edgar also mentioned this in Santiago.

Although the use case for identifying the known drugs that are associated with targets can be solved (drug_type filter). It is available in Target Info from DrugBank ?db_compound drugbank:genericName ?drug_name ; drugbank:drugType ?drugType ;

On Feb 20, 2015, at 4:26 PM, danidi notifications@github.com<mailto:notifications@github.com> wrote:

In ChEMBL, the connection of a drug compound to it's target is available via the drug_mechanism field. Currently, the use-case identifying targets for known drugs is not possible with the system. Either adding the drug mechanism information (e.g. to Compound Information), or the drugbank information (see #233 https://github.com/openphacts/GLOBAL/issues/233), or both, would really enhance the system.

— Reply to this email directly or view it on GitHub https://github.com/openphacts/GLOBAL/issues/235.

— Reply to this email directly or view it on GitHubhttps://github.com/openphacts/GLOBAL/issues/235#issuecomment-75423406.

[NuriaQueralt]

Hi all,

I agree that exposing this information in a harmonised way would be really relevant. According what Anna Gaulton mentioned in a previous discussion regarding that, it seems that BAO ontology could align fairly well at least with ChEMBL. Also, there is the Functional Therapeutic Chemical Classification from EBI that may be of interest. Underlying FTC there are DrugBank and GO/GOA annotations:

BAO: http://bioportal.bioontology.org/ontologies/BAO Looking for 'Agonist': http://bioportal.bioontology.org/ontologies/BAO/?p=classes&conceptid=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FCHEBI_48705&jump_to_nav=true

FTC: http://chembl.blogspot.com.es/2013/11/paper-functional-therapeutic-chemical.html Looking for 'Agonist': https://www.ebi.ac.uk/chembl/ftc/search/agonist

@annagaulton answer about ChEMBL: We model this in two different ways in ChEMBL (since we have two different types of association). The cco:hasMechanism predicate is used to describe the known mechanism of action of approved drugs i.e., the target believed to be responsible for their efficacy in a disease. This information has been included in the RDF since chembl_18. But we also have compound to target associations via assays that have been reported in literature. These are modelled as follows (i.e., a molecule has an activity in an assay which has a target):

cco:Molecule cco:hasActivity cco:Activity cco:Activity cco:hasAssay cco:Assay cco:Assay cco:hasTarget cco:Target

We created a ChEMBL Core Ontology (CCO) to model the ChEMBL RDF, based on the underlying ChEMBL data model (https://www.ebi.ac.uk/rdf/documentation/chembl). We looked at other ontologies but there wasn’t one that covered all aspects (BioAssay Ontology is probably closest) and we wanted to keep the model recognisable for existing ChEMBL users. We had in mind that if we wanted to align the model with other resources we could create mappings between the CCO terms and other ontologies.

I can’t recall the exact way DrugBank is modelled (maybe someone else can advise on that), but it is different from ChEMBL (I think they also just have their own predicates) and any alignment would just be at the level of the API queries I guess. Initially the type of association was a little different from ChEMBL because DrugBank has drug mechanism of action data and we only had bioactivity/assay data (the ChEMBL mechanism of action stuff is not yet in OPS production system), so we haven’t really tackled that issue yet I think.

Cheers, n

On 02/22/2015 10:09 AM, Chris Evelo wrote:

Yes, absolutely.

But I suppose we will get the information automatically when we update to the latest ChEMBL and Drugbank versions. The question would be how to harmonise that information across the two resources and how we expose that again on the API. Any ideas on that?

Best, Chris

— Prof.Dr. Chris Evelo Dept. Head Bioinformatics – BiGCaT Maastricht University

From: ChristineChichester notifications@github.com<mailto:notifications@github.com> Reply-To: openphacts/GLOBAL reply@reply.github.com<mailto:reply@reply.github.com> Date: Sunday 22 February 2015 8:13 To: openphacts/GLOBAL GLOBAL@noreply.github.com<mailto:GLOBAL@noreply.github.com> Subject: Re: [GLOBAL] Add Drug Mechanism of Action information from ChEMBL (#235)

The mechanism of action would be a nice addition. Edgar also mentioned this in Santiago.

Although the use case for identifying the known drugs that are associated with targets can be solved (drug_type filter). It is available in Target Info from DrugBank ?db_compound drugbank:genericName ?drug_name ; drugbank:drugType ?drugType ;

On Feb 20, 2015, at 4:26 PM, danidi notifications@github.com<mailto:notifications@github.com> wrote:

In ChEMBL, the connection of a drug compound to it's target is available via the drug_mechanism field. Currently, the use-case identifying targets for known drugs is not possible with the system. Either adding the drug mechanism information (e.g. to Compound Information), or the drugbank information (see #233 https://github.com/openphacts/GLOBAL/issues/233), or both, would really enhance the system.

— Reply to this email directly or view it on GitHub https://github.com/openphacts/GLOBAL/issues/235.

— Reply to this email directly or view it on GitHubhttps://github.com/openphacts/GLOBAL/issues/235#issuecomment-75423406.

— Reply to this email directly or view it on GitHub https://github.com/openphacts/GLOBAL/issues/235#issuecomment-75426576.Web Bug from https://github.com/notifications/beacon/ACIU1YMEJfMm41M5497Ge-iFBTq4fVSPks5nuZRBgaJpZM4DjWde.gif

---

Núria Queralt Rosinach Research Programme on Biomedical Informatics (GRIB) Department of Experimental and Health Sciences Universitat Pompeu Fabra IMIM (Hospital del Mar Medical Research Institute) C/Dr. Aiguader 88, 08003 Barcelona, Spain Tel.: +34 93 316 0536 (1536) E-mail: nqueralt@imim.es Skype IM: nuriaqr76 http://ibi.imim.es/

---

[danidi]

I think one problem with harmonizing is that we always first mention the datasource. So there would be two different blocks of data, one for drugbank and the other for chembl where basically the same information could be retrieved. Maybe @antonisloizou can comment on that.

@ChristineChichester Yes, target to compound is no problem. It's just the other way round that is missing.

[NuriaQueralt]

Hi,

Just to notice that in the eTOX project they are implementing a MoA ontology that it seems that it is based on the ChEMBL one. I think the eTOX MoA ontology would be an interesting option for Open PHACTS. This would be an opportunity for both projects to work in synergy and to avoid redundant efforts, and to foster data standardization and data sharing.

Cheers, n

[mark-davies]

@NuriaQueralt - Do you have a link to the eTOX MoA work?

Cheers,

Mark

[NuriaQueralt]

@mark-davies the access to the eTOX MoA ontology will be available as open this month. Nevertheless, let me ask them for a link that i could send you.

Cheers, Núria

[NuriaQueralt]

Hi all,

@ChristineChichester , The eTOX MoA ontology developers finally answered me! This is what they said to me:

'the MoA ontology was created based on the vocabulary which is included in the eTOX database.

It’s not the whole ontology with all targets like this one from Chembl. I’m currently comparing the etox targets in order to check how many new targets we have which are not included in the Chembl list.

The etox MoA ontology needs a lot of more work, because of an amount of unclassified vocabulary and new endpoints (because new database release) – it’s not finished at all.'

If you are interested to look at the ontology i can ask for the OWL file.

@mark-davies, the contact for this work in ChEMBL is Patricia Bento (patricia@ebi.ac.uk), if you are interested.

Also, i could met the MoA developers here in Barcelona in 4-7 of May to show me the work that they have done til the moment.

Cheers, Núria