Roche

#Goal: Graph pattern for drug efficacy in knowledge graph with semantic data#

This use case is based on the following paper:

http://rdcu.be/khyn

In this paper a protein-protein interaction network was used, which lacks semantics such as explicit predicates and sources for the relationships.

The goal of the Roche use case to recreate this experiment using the ODEX Knowledge Platform (OKP), while using the same input data of the Nature paper.

Machine learning will be applied to the extracted data to extract patterns from the OKP. We have not yet reached a final decision how this extracted data will be represented, nor which machine learning algorithm will be used. However, Anand suggested the Caret package for R, which cycles through ML algorithms and determines which is best.

Aram (Euretos) has supplied us with a positive set of efficacious drugs. A set of negative signals (non-efficacious drugs) still needs to be collected, for which Anneke has suggested a number of resources.

_"Clinical Candidate Database (CCD) – Contains compounds that either are or were in various clinical phases and captured from over 40,000 Journal articles and other publicly available data sources. Clinical Candidate Database consists of chemically diverse compounds/records with Biological, Pharmacological, Clinical pharmacokinetic Information which are in various stages of drug development. Besides it also includes suspended, discontinued drugs which are/were failed in different phases. Data is extracted from the articles published in various standard journals covering clinical development of new drugs and also from patents. Clinical Candidate Database is available in different user-friendly searchable formats such as ISIS/Base, SD-format, RD-format and Oracle Dump. _ → zit in OKP, maar weet niet of de info die je nodig hebt ook in OKP zit...

Deze referentie? Minna Allarakhia: Open-source approaches for the repurposing of existing or failed candidate drugs: learning from and applying the lessons across diseases. Drug Des Devel Ther. 2013; 7: 753–766. doi: 10.2147/DDDT.S46289. PMCID: PMC3743608

_http://www.pharmgkb.org drugs that fail or vary in effectiveness due to human genetic variations _ → voegt wel een extra nivo van complexiteit toe

http://www.fda.gov/Drugs/InformationOnDrugs/ucm079750.htm This is a great resource. It will let you download a database of FDA-approved drugs that have since been withdrawn. You can filter ProductMktStatus=Discontinued.

Wikipedia has a decent list of withdrawn drugs http://en.wikipedia.org/wiki/List_of_withdrawn_drugs"