Successfully reported this slideshow.
Your SlideShare is downloading. ×

Open PHACTS webinar June 2016 - Data2Discovery

Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Upcoming SlideShare
Pharma data analytics
Pharma data analytics
Loading in …3
×

Check these out next

1 of 18 Ad

More Related Content

Slideshows for you (20)

Advertisement

Similar to Open PHACTS webinar June 2016 - Data2Discovery (20)

More from open_phacts (18)

Advertisement

Recently uploaded (20)

Open PHACTS webinar June 2016 - Data2Discovery

  1. 1. DATA2DISCOVERY OpenPHACTS: Maximizing Impact for Pharmaceutical Applications DAVID WILD, PhD. CEO Data2Discovery Inc Associate Professor and Director of Data Science, Indiana University http://d2discovery.com - http://djwild.info - david@d2discovery.com DATA SCIENCE DATA2DISCOVERY 1
  2. 2. DATA2DISCOVERY Data2Discovery Inc •  Finding insights from proprietary, public and commercial data together in ways never tried before •  Linking data from molecular to human •  Key expertise in using OpenPHACTS data •  Partnering with customers to implement this at scale •  Phenotypic drug discovery is a beachhead application •  http://d2discovery.com 2
  3. 3. DATA2DISCOVERY Data2Discovery Application Areas Pharmaceutical Research Healthcare Delivery Public Health Precision Medicine Population Health Real World Evidence Combination therapy Outcome Analytics Nudge Economics Quantified Self Performance-based Pricing Risk Adjustment Phenotypic Drug Discovery Adverse Event Prediction 3
  4. 4. DATA2DISCOVERY IT Infrastructure Re-Imagined 4 What effect will inhibiting PPAR-γ have on HER+ breast cancer tissue? Interaction Infrastructure (II) Custom interfaces and applications built using reusable components Computation Infrastructure (CI) Fast, in memory computation modules that work on heterogeneous data sub-graphs Linked Data Ecosystem (LDE) Data APIs, MicroTranslators, Live Triple Stores, Ontologies & Standards What adverse effects could Cymbalta cause in patients with Chron’s?
  5. 5. DATA2DISCOVERY Case study – Eli Lilly •  Linked data ecosystem: Developed live Data Translator that maps key internal assay data into OpenPHACTS triple store, annotated by public ontologies (e.g. BAO) •  Computation/Interaction infrastructure: Contextual inquiry in high impact end-user applications across the organization 5
  6. 6. DATA2DISCOVERY Method Target UniProt ChEMBL Assay Internal Assay Data ChEMBL Compd Semantic Schema Assay Result Type ChEMBL Cell Line Gene Cell Line Mapping Internal and External Assay Data 6
  7. 7. DATA2DISCOVERY Phenotypic Drug Discovery •  “Beachhead” application •  Subject of strong Pharma interest (OpenPHACTS researchathon Spring 2015) •  Crosses the big two domains –  Chemistry & biology (molecular) –  Study of people (patient) •  Need to find links across many datasets –  Preclinical, toxicology, clinical trials, post-market 7
  8. 8. DATA2DISCOVERY P3 – Predictive Phenotypic Profiler •  P3 is a product in development by Data2Discovery that is designed to address current gaps in maximizing impact of phenotypic assay data. Intended applications include: –  Target deconvolution –  Target-based mechanism of action discovery –  Identifying similarities between phenotypic assays •  These use cases were identified in OpenPHACTS Phenotypic Screening Workshop in February 2015 •  Identifies associations using SEMAPTM association finding technology •  Links OpenPHACTS data into other key data resources 8
  9. 9. DATA2DISCOVERY P3 Architecture •  Componentized architecture –  Permits plug-and-play of datasets –  Allows new applications to be quickly developed •  Highly scalable –  Uses cloud technology, fast graph databases and Apache SPARK •  Secure deployments where needed –  Can be used as external software-as service –  Can be deployed behind firewalls (Docker images) 9
  10. 10. DATA2DISCOVERY What types of data can P3 use? •  P3 can link public and proprietary data sources and cross pre- clinical and clinical data. Example types of data include: –  Enzymatic assay (compound-gene/target) –  Phenotypic assay (compound-phenotype) –  Cellular assay (compound-cell line) –  Gene expression (e.g. LINCS, L1000) –  Pathway (pathway-gene) –  Molecular-Phenotype links (compound, gene, pathway, cell line to disease state, genotype, patient phenotype, adverse event, electronic medical record, real world evidence etc) 10
  11. 11. DATA2DISCOVERY SEMAP™ - Semantic Association Prediction •  Based on research at Indiana University •  Predicts association based on data subnetworks between points of interest •  Can predict drug-target interaction across thousands of genes •  No scope or bias problems •  Extensive external validation including comparison with SEA and use for predicting gene expression profiling (see Chen, B. et al., PLoS Computational Biology, 2012, 8(7), e1002574) •  Demonstrated applications in drug repositioning, MOA discovery •  Now being used in phenotypic target deconvolution and MOA discovery 11
  12. 12. DATA2DISCOVERY Example: Troglitazone and PPARG Association score: 2385.9 Association significance: 9.06 x 10-6 => missing link predicted 12
  13. 13. DATA2DISCOVERY P3 Public Data Demonstration •  Links several public datasets using TB as a sample TA: •  ChEMBL/OpenPHACTS (enzymatic, cellular assays) •  NCATS Phenotypic Drug Discovery Resource1 (phenotypic assays) •  Manual phenotypic assay – pathway/gene annotations •  TB drug – gene (DrugBank, NCATS) 1https://ncats.nih.gov/expertise/preclinical/pd2 13
  14. 14. DATA2DISCOVERY Open Phenotypic Drug Discovery Resource Eli Lilly IT https://ncats.nih.gov/expertise/preclinical/pd2 14
  15. 15. DATA2DISCOVERY Current P3 Linked Data Network 15
  16. 16. DATA2DISCOVERY P3 Public Data Example 16
  17. 17. DATA2DISCOVERY Drug Discovery Development & Clinical Trials Post Market /Healthcare Use Product Example Data Open PHACTS OPDDR ToxCast Internal Cpd/Assay Internal Trials Data ClinicalTrials.gov Quantified Self / IOT Medicare CMS EMR Data RWE Data P3-Discovery P3-Clinical P3-PopHealth Target ID Poly- Pharmacology MOA Discovery ADME/ Tox Adverse Event Prediction Candidate Repositioning Drug Repositioning Population Analysis Drug Combinations Personalized Medicine FAERS Value/Performance based Pricing P3-Development Plan 17
  18. 18. DATA2DISCOVERY What can Data2Discovery offer? •  Agile development of high impact applications that use semantic linked data •  Implementation of linked data ecosystem, computation infrastructure and user components •  Key expertise in using OpenPHACTS data •  Contact David Wild, david@d2discovery.com 18

×