Addressing gaps in clinically useful evidence on potential drug-drug interactions
-
Upload
richard-boyce-phd -
Category
Sports
-
view
441 -
download
1
description
Transcript of Addressing gaps in clinically useful evidence on potential drug-drug interactions
Biomedical Informatics1
Addressing gaps in clinically useful evidence on drug-drug interactionsMay 2nd 2013 BioDLP Seminar at the University of Wisconsin - Milwaukee
Richard Boyce, University of PittsburghDepartment of Biomedical Informatics
Biomedical Informatics2
Goals for this talk• Describe potential drug-drug
interactions (PDDIs)– the significant challenges facing clinicians and
maintainers of drug information systems.
• Present a new PDDI knowledge representation paradigm– that I hypothesize will yield more clinically
relevant evidence than is currently possible
• Discuss my BioDLP research– Within the context of the new paradigm
Biomedical Informatics3
Part I – PDDIs and challenges for PDDI knowledge representation
Biomedical Informatics4
What is a PDDI?
• Drug-drug interaction:– a clinically meaningful alteration of the
effect of a drug (object drug) occurs as a result of coadministration of another drug (precipitant drug) [10]
• Potential drug-drug interaction (PDDI):– two drugs known to interact are prescribed
whether or not harm ensues [10]
•Pharmacokinetic or pharmacodynamic
Biomedical Informatics5
The clinical importance of PDDIs• PDDIs are a significant source of
preventable drug-related harm– 13.3% of preventable errors leading to
an ADE [1]
– 7% (23/338) of the ADEs attributable to PDDIs [2]
– 16 cohort and case-control studies reported an elevated risk of hospitalization in elderly patients who were exposed to PDDIs [3]
Biomedical Informatics6
Knowledge is important• Failure to properly manage a PDDI is
a medical error• The IOM has noted that a lack of
drug knowledge is one of the most frequent proximal causes such errors [4]
Biomedical Informatics7
The danger of incomplete drug-drug interaction knowledge
Biomedical Informatics8
Key point
• Many drug information systems disagree about PDDIs– the specific ones that exist
– their potential to cause harm
• This leads to– confusion and frustration for clinicians
– greater risks of harm to patients
Biomedical Informatics9
Evidence of drug compendia problems• Three PDDI information sources agreed
upon only 25% of 59 contraindicated drug pairs found in black box warnings [5]
• 18 (28%) of 64 pharmacy information and clinical decisions support systems correctly identified 13 clinically significant DDIs [6]
• Four sources agreed on only 2.2% of 406 PDDIs considered to be “major” by at least one source [7]
Biomedical Informatics10
Why do compendia disagree?• Four types of information to decide if a PDDI
warrants clinical action. [21]
• Collecting evidence related to each information item on 244 PDDIs enabled them to determine that 12% would require no action by physicians [8]
A conceptual model – 30,000 feet view
Limits the effectiveness of
PDDI alerting and CPOE systems
Drug Compendia synthesize PDDI evidence into knowledge but•May fail to include important PDDIs•Often disagree about PDDI evidence and seriousness ranking•May include numerous PDDIs with little evidence for liability reasons
PDDI adverse event
Increases the risk of
PDDI evidence
Scattered across numerous sources
Biomedical Informatics12
PDDI evidence – pre-market studies
Pre-market studies establish PDDI feasibility but:•usually do not indicate ADE seriousness, incidence, or risk•Focus on generally younger and healthier populations•Do not exist for many older drugs
Product labeling
Reported in
Scientific literature
Rarely reported in
See references 31 and 32
Biomedical Informatics13
PDDI evidence – post-market studies
Post-market studies can provide evidence of PDDI risk and incidence if well-designed but:•rarely are randomized studies due to ethical considerations•older drugs less likely to be studied
Product labeling
Scientific literature
Reported inRarely reported in
Biomedical Informatics14
PDDI evidence – Clinical experience
Product labeling
Clinical experience can provide first warning of a PDDI's and offers unique insight on PDDI severity:•are often case reports of low evidential quality•there is no general way to collect and share these insights
Rarely reported inRarely reported in
Scientific literature
Evidence from the drug compendium perspective
Pre-market studies
Post-market studies
Product labeling
Reported in
Clinical experience
Scientific literature
Rarely reported in
Rarely reported inReported in
Rarely reported in
Drug Compendia synthesize PDDI evidence into knowledge but•May fail to include important PDDIs•Often disagree about PDDI evidence and seriousness ranking•May include numerous PDDIs with little evidence for liability reasons
Source forSource for
Biomedical Informatics16
Effects on the clinician and patient
PDDI alerting and CPOE systems
Drug Compendia synthesize PDDI evidence into knowledge but•May fail to include important PDDIs•Often disagree about PDDI evidence and seriousness ranking•May include numerous PDDIs with little evidence for liability reasons
PDDI adverse event
Increases the risk of
Limits the effectiveness of
Biomedical Informatics17
PDDI over-alerting
• Systems that provide PDDI alerts at the point of care often alert to PDDIs that have little potential clinical significance– frustrating clinicians
“Drug safety alerts are overridden by clinicians in 49% to 96% of cases” [11]
– can lead to inappropriate responses
“An increased number of non-critical alerts…was the only variable associated with an inappropriate provider response” [12]
Biomedical Informatics18
Summary of challenges for PDDI knowledge representation• PDDI evidence is distributed, dynamic,
and of varying quality• There are significant gaps in PDDI
evidence making it hard to assess – what is the potential harmful effect?
– who is the PDDI most likely to affect?
– when is a patient most at risk?
• Alerting has to become more intelligent!
Biomedical Informatics19
Part II – a new PDDI knowledge representation paradigm
The new paradigmProduct labeling
Scientific literature
A framework for representing PDDI assertions as interoperable Linked Data
Pharmacoepidemiology studies
Semantic annotation
High priority PDDIs for research
Semantic annotation
Reduced risk of a PDDI
medication error!
Clinical experience
Better synthesis of PDDI evidence, easier identification of gaps
Expected benefits:•More complete and accurate PDDI evidence•Better informed pharmacists and other clinicians•More effective PDDI alerting and decisions support systems
Biomedical Informatics21
Elements of the new paradigm• Linked Data [13]
– a Semantic Web technology that makes distributed knowledge sources interoperable, with interconnections providing rich context that would be unavailable from any single database
• Semantic annotation [14]– a technology that enhances digital
information artifacts by linking them to provenance and expert commentary
• Pharmacoepidemiology [15]– an approach to studying of the use and
effects of drugs in large numbers of people
Biomedical Informatics22
Linked Data• What is it?
– 3 minute jargon free introduction: • player.vimeo.com/video/36752317
• My research has shown Linked Data to be a potentially effective means of linking clinical drug information [9]– Several high quality resources
– More complete information
Biomedical Informatics23
predicate
Resource Description Framework (RDF)
• Data model – triples
• Syntax – RDF– The subject, predicate, and objects are specified by
URIs<http://.../AnneHathaway> <http://.../Married> <http://../Shakespeare>
<http://.../Shakespeare> <http://.../Wrote> <http://../Hamlet>
subject object
AnnHathawayShakespeare
Hamlet
marriedwrote
Biomedical Informatics24
Semantic Annotation
http://www.openannotation.org/spec/core/
Semantic Annotation of PDDIs
Combining Linked Data and Semantic Annotation
Biomedical Informatics27
A structured assessment scores evidence and potential severity [21]
Pharmacoepidemiology – filling in the gaps
Recap of the new paradigmProduct labeling
Scientific literature
A framework for representing PDDI assertions as interoperable Linked Data
Pharmacoepidemiology studies
Semantic annotation
High priority PDDIs for research
Semantic annotation
Reduced risk of a PDDI
medication error!
Clinical experience
Better synthesis of PDDI evidence, easier identification of gaps
Expected benefits:•More complete and accurate PDDI evidence•Better informed pharmacists and other clinicians•More effective PDDI alerting and decisions support systems
Biomedical Informatics29
Anticipated benefits of the new paradigm• A computable representation of PDDI safety
concerns that is linked to:– evidence
– expert input, and
– pharmacoepidemiologic study results
• More complete, timely, and accurate PDDI evidence – easier integration for drug compendia and CPOE
developers
• Better informed clinicians and patients
Biomedical Informatics30
Part II – A brief review of my research within the context of this paradigm
Overview of my recent PDDI studies
Product labelingScientific literature
A framework for representing PDDI assertions as interoperable Linked Data
Pharmacoepidemiology studies
Semantic annotation
High priority PDDIs for research
Semantic annotation
Clinical experience
Better synthesis of PDDI evidence, easier identification of gaps
A, B
C
E G
A. Boyce et al. Am J Geriatr Pharmacother. 2012. Apr;10(2):139-50. [22] B. Boyce et al. Annals of Pharmacotherapy. 2012. Oct;46(10):1287-98 [23] C. Boyce et al. Proceedings of the 2012. Workshop on BioNLP. 206-213 [16] D. & E. Boyce et al. Proceedings of the 2013. AMIA Summit on Translational Bioinformatics. 28-32 (D), 64-68 (E). [18,19]F. Boyce et al. J Biomed Semantics. 2013. Jan 26;4(1):5. [9]G. Boyce et al. Poster at Aging Institute Research Day. 2013. [20]
F
D
Linked Data – linking product labels to the “Web of Drug Identity”
Product labeling
A framework for representing PDDI assertions as interoperable Linked Data
C. Boyce et al. Proceedings of the 2012. Workshop on BioNLP. 206-213 [16] E. Boyce et al. Proceedings of the 2013. AMIA Summit on Translational Bioinformatics. 64-68 [19]F. Boyce et al. J Biomed Semantics. 2013. Jan 26;4(1):5. [9]
E
Hypothesis: A Linked Data knowledge base of drug product labels with accurate links to other relevant sources of drug information will provide a dynamic platform for drug information NLP that provides real value to clinical and translational researchers.
Better synthesis of PDDI evidence, easier identification of gaps F
C
Biomedical Informatics33
Structured Product Labels (SPLs)
• All package inserts for currently marketed drugs are available as SPLs [27-29]
Biomedical Informatics34
More about SPLs
Biomedical Informatics35
Key point
• LinkedSPLs [26] is a Linked Data version of SPLs– >36,000 FDA-approved prescription and
over-the-counter drugs present in DailyMed
– simplifies access to SPL content
– interoperable with other important drug terminologies and resources
– Enables queries across drug information resources…
Biomedical Informatics36
Example cross-resource queries• What are the known targets of all active
ingredients that are classified as antidepressants?
• Is there a pharmacogenomics concern for any of the drugs associated with Hyperkalemia
• Show the evidence support for all pharmacokinetic PDDIs affecting buproprion that are supported by a randomized study
Biomedical Informatics37
LinkedSPLs – A research program
Biomedical Informatics38
LinkedSPLs – A research program
Your annotations would go here!
Biomedical Informatics39
An Example - extracting PDDIs from product labelsProduct labeling
C
C. Boyce et al. Proceedings of the 2012. Workshop on BioNLP. 206-213 [16]
Recently published NLP algorithm specifically designed to extract PDDIs from drug product labels
• A drug package insert PK PDDI corpus [24]:• 592 PK PDDIs, • 3,351 active ingredients, • 234 drug product mentions, • 201 metabolite mentions.
• SVM performed best • F = 0.859 for pharmacokinetic PDDI identification • F = 0.949 for modality assignment
• Syntactic information helped with sentences containing both interacting and non-interacting pairs
Biomedical Informatics40
Application of the NLP algorithm• Extract PDDIs and integrate into linked
SPLs
PDDI Extraction algorithm
Lovastatin product label
Biomedical Informatics41
Linkage to external sources• There are many sources of drug information
that are complementary to each other.– DrugBank: contains drug targets, pathways,
interactions
– RxNorm: provides UMLS mappings
– VA NDF-RT: PDDIs and drug classification
– ChEBI: provides rigorous classification of drugs
Biomedical Informatics42
Two linking studies
• Active ingredients in the structured portion of SPLs to DrugBank [19]• Three different approaches
• One fully unsupervised
• PDDIs (VA NDF-RT) to the Drug Interactions section of 26 psychotropics [9]• What benefits for this linkage?• Collaboration with Majid Rastegar-Mojarad
Biomedical Informatics43
Linking Active ingredients in SPLs to DrugBank• Three different linking approaches to link
to DrugBank1. Structure string (InChI)
2. Ontology label matching (ChEBI)
3. Unsupervised linkage point discovery (Automated) [30]
Biomedical Informatics44
Linkage to DrugBank – Results
• 1,246 active ingredients (53%) could be mapped to DrugBank by at least one method
• 1,096 unmapped ingredients
• The three approaches complement each other
InChI identifier
ChEBI identifier
InChI + ChEBI
Automatic
InChI identifier 424 261 424 395
ChEBI identifier --- 707 707 650
InChI + ChEBI -- -- 831 791
Automatic -- -- -- 1162
Biomedical Informatics45
• The automatic approach performs very well– A greater number of accurate links discovered
with less effort
• A significant number remain unmapped:– Some salt or racemic forms of mapped
ingredients (e.g., alpha tocopherol acetate D)– Elements (e.g., gold, iodine), and variety of
natural organic compounds including pollens (N~200)
• Not all ingredients are included in DrugBank– other resources may be required to obtain
complete mappings for active ingredients.
Linking methods conclusions
Biomedical Informatics46
Linking from VA NDF-RT to LinkedSPLs• How often would it provide more
complete information?VA NDF-RT in Bioportal
October 2012
SPARQL: Get all DDIs for
antidepressants
Filter out DDIs previously
identified in antidepressant product labels
Tabulate potentially novel
PDDIs
Biomedical Informatics47
PDDI studies comparing information sources
Product label PDDIs for 20 drugs manually identified [22]• ~70 interactions • Pharmacokinetic and pharmacodynamic
We filtered NDF-RT PDDIs • String matching and an expanded version of the PDDI table
• ~2,500 drug-drug and drug-class pairs
Face validity but future work needed for • validate the accuracy of this approach• create a more scalable approach
Filter out DDIs previously
identified in antidepressant product labels
Biomedical Informatics48
Linking from VA NDF-RT - results• At least one potentially novel interaction was linked to
a product label for products containing each of the 20 antidepressants– tranylcypromine (33), nefazodone (31), fluoxetine (28)
• Several cases where all of the PDDIs were potentially novel– e.g., trazodone, venlafaxine, trimipramine
• Pharmacist review– Several true positives
• e.g., escitalopram-tapentadol, escitalopram-metoclopramide
– Some false positives
• e.g., nefazodone-digoxin (digitalis)
Biomedical Informatics49
Proof of concept mashup
Biomedical Informatics50
The complete proof of concept
• http://tinyurl.com/c53skm4 – 29 psychotropic drug products – Created using four Semantic Web nodes
1. http://thedatahub.org/dataset/linked-structured-product-labels2. http://thedatahub.org/dataset/linkedct3. http://bioportal.bioontology.org/ontologies/471014. http://thedatahub.org/dataset/the-drug-interaction-knowledge-base
Biomedical Informatics51
Concluding points
• The paradigm provides a framework for tying together– NLP for extracting PDDIs
– NLP for linking evidence it to PDDIs
– Aggregating existing PDDI resources in a single framework• Research prioritization
• Community engagement
Biomedical Informatics52
Acknowledgements• Agency for Healthcare Research and Quality
(K12HS019461). • NIH/NCATS (KL2TR000146),• NIH/NIGMS (U19 GM61388; the Pharmacogenomic
Research Network) • NIH/NLM (T15 LM007059-24)• The Drug Interaction Knowledge Base team
– John Horn Pharm.D, Carol Collins MD, Greg Gardner, Rob Guzman
• W3C LODD Task Force and the Clinical Genomics Working Group
Biomedical Informatics53
References1. Gurwitz JH, Field TS, Harrold LR, et al. Incidence and preventability of
adverse drug events among older persons in the ambulatory setting. JAMA. 2003;289(9):1107–1116
2. Gurwitz JH, Field TS, Judge J, et al. The incidence of adverse drug events in two large academic long-term care facilities. Am. J. Med. 2005;118(3):251–258
3. Hines LE, Murphy JE. Potentially harmful drug-drug interactions in the elderly: a review. Am J Geriatr Pharmacother. 2011;9(6):364–377.
4. Committee on Identifying and Preventing Medication Errors, Philip Aspden, Julie Wolcott, J. Lyle Bootman, Linda R. Cronenwett, Editors. Preventing Medication Errors: Quality Chasm Series. Washington, D.C.: The National Academies Press; 2007.
5. Wang LM, Wong M, Lightwood JM, Cheng CM. Black box warning contraindicated comedications: concordance among three major drug interaction screening programs. Ann Pharmacother. 2010;44(1):28–34
6. Saverno KR, Hines LE, Warholak TL, et al. Ability of pharmacy clinical decision-support software to alert users about clinically important drug-drug interactions. J Am Med Inform Assoc. 2011;18(1):32–37.
Biomedical Informatics54
References cont.7. Abarca J, Malone DC, Armstrong EP, et al. Concordance of severity ratings provided in four drug interaction compendia. J Am Pharm Assoc (2003). 2004;44(2):136–141.
8. Van Roon EN, Flikweert S, Le Comte M, et al. Clinical relevance of drug-drug interactions : a structured assessment procedure. Drug Saf. 2005;28(12):1131–1139.
9. Boyce R, Horn J, Hassanzadeh O, et al. Dynamic Enhancement of Drug Product Labels to Support Drug Safety, Efficacy, and Effectiveness. Journal of Biomedical Semantics. Journal of Biomedical Semantics. 2013. Jan 26;4(1):5.
10. Hines LE, Malone DC, Murphy JE. Recommendations for Generating, Evaluating, and Implementing Drug-Drug Interaction Evidence. Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy. 2012;32(4):304–313.
11. Van der Sijs H, Aarts J, Vulto A, Berg M. Overriding of drug safety alerts in computerized physician order entry. J Am Med Inform Assoc. 2006;13(2):138–147.
Biomedical Informatics55
References cont.12. Miller AM, Boro MS, Korman NE, Davoren JB. Provider and pharmacist responses to warfarin drug-drug interaction alerts: a study of healthcare downstream of CPOE alerts. J Am Med Inform Assoc. 2011;18 Suppl 1:i45–50. PMCID: PMC3241165
13. Marshall MS, Boyce RD, Deus H, Zhao J, Willighagen E, Samwald M, Pichler E, Hajagos J, Prud’hommeaux E, and Stephens, S. Emerging practices for mapping life sciences data to RDF - a case series. Journal of Web Semantics. Special Issue: Reasoning with Context in the Semantic Web. Volume 14, July 2012, Pages 2–13.
14. Open Annotation Collaboration. http://www.openannotation.org/
15. Strom BL, Kimmel SE eds. Textbook of Pharmacoepidemiology. 1st ed. Wiley; 200716. Boyce R, Gardner G, Harkema H. Using Natural Language Processing to Extract Drug-Drug Interaction Information from Package Inserts. Proceedings of the 2012 Workshop on BioNLP. Montreal, Quebec, Canada. June 2012:206-213. https://www.aclweb.org/anthology/W/W12/W12-2426.pdf
Biomedical Informatics56
References cont.17. Rasteger-Mojarad, M., Boyce RD., Prasad, R. UWM-TRIADS: Classifying Drug-Drug Interactions with Two-Stage SVM and Post-Processing. Proceedings of the 2013 International Workshop on Semantic Evaluation (SemEval), Task 9 - Extraction of Drug-drug Interactions from BioMedical Texts. Atlanta Georgia, June 2013. (In Press).
18. Boyce, RD., Freimuth, RR., Romagnoli, KM., Pummer, T., Hochheiser, H., Empey, PE. Toward semantic modeling of pharmacogenomic knowledge for clinical and translational decision support. Proceedings of the 2013 AMIA Summit on Translational Bioinformatics. San Francisco, March 2013:28-32.
19. Hassanzadeh, O., Zhu, Qian., Freimuth, RR., Boyce R. Extending the “Web of Drug Identity” with Knowledge Extracted from United States Product Labels. Proceedings of the 2013 AMIA Summit on Translational Bioinformatics. San Francisco, March 2013:64-68.
20. Boyce RD., Handler SM., Karp JF., Perera, S., Hanlon JT. Prevalence of Potential Drug-Drug Interactions Affecting Antidepressant in US Nursing Home Residents. Poster presentation at the 2013 Aging Institute Research Day. University of Pittsburgh. Pittsburgh PA. April, 2013
Biomedical Informatics57
References cont.21. E. N. van Roon, S. Flikweert, M. le Comte, P. N. Langendijk, W. J. Kwee-Zuiderwijk, P. Smits, and J. R. Brouwers. Clinical relevance of drug-drug interactions : a structured assessment procedure. Drug Saf, 28(12):1131-1139, 2005.
22. Boyce RD, Handler SM, Karp JF, Hanlon JT. Age-related changes in antidepressant pharmacokinetics and potential drug-drug interactions: a comparison of evidence-based literature and package insert information. Am J Geriatr Pharmacother. 2012 Apr;10(2):139-50. Epub 2012 Jan 27. PMID 22285509. PMCID: PMC3384538
23. Boyce RD, Collins C, Clayton M, Kloke J, Horn J. Inhibitory Metabolic Drug Interactions with Newer Psychotropic Drugs: Inclusion in Package Inserts and Influences of Concurrence in Drug Interaction Screening Software. Annals of Pharmacotherapy. 2012 Oct;46(10):1287-98. Epub 2012 Oct 2. DOI 10.1345/aph.1R150. PMID 23032655
24. http://purl.org/NET/nlprepository/PI-PK-DDI-Corpus
25. http://www.openannotation.org/spec/core/
26. http://purl.org/net/linkedspls
Biomedical Informatics58
References cont.27. http://www.fda.gov/OHRMS/DOCKETS/98fr/FDA-2005-N-0464-gdl.pdf
28. http://goo.gl/C8Vv2
29. http://dailymed.nlm.nih.gov/dailymed/downloadLabels.cfm
30. O. Hassanzadeh et al. “Discovering Linkage Points over Web Data”. To Appear in PVLDB, Vol 6. Issue 6, August 2013
31. FDA. Guidance for Industry Drug Interaction Studies — Study Design, Data Analysis, Implications for Dosing, and Labeling Recommendations. Silver Spring, MD: Federal Drug Administration; 2012. Available at: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm292362.pdf. Accessed January 7, 2013.
32. Platt R, Wilson M, Chan KA, Benner JS, Marchibroda J, McClellan M. The new Sentinel Network--improving the evidence of medical-product safety. N Engl J Med. 2009 Aug 13;361(7):645-7.
Biomedical Informatics59
Backup Slides
Semantic annotation and Linked Data
Product labeling
A framework for representing PDDI assertions as interoperable Linked Data
Semantic annotation
D. Boyce et al. Proceedings of the 2013. AMIA Summit on Translational Bioinformatics. 28-32 [18]
D
D – Semantic annotation of pharmacogenomics statements in drug product labeling
• First semantically annotated corpus of clinical pharmacogenomics statements present in drug product labeling
• Potential impact• pharmacokinetic / pharmacodynamic
• Patient specific risk factors• concomitant medications• medical conditions
• Recommendations• dosage, drug administration, alternatives, monitoring, and tests
• First pharmacogenomics dataset to use the W3C Open Data Annotation standard [25]
Biomedical Informatics61
Pharmacogenomics example - Codeine
Predicate Object
drug CODEINE
biomarker CYP2D6
variant Ultra-rapid metabolizer
Pharmacokinetic effect
Metabolism-increase
Pharmacodynamic effect
Drug-toxicity-risk-increase
Predicate Object
hasSource URL to product label
Exact-text “Nursing mothers…”
Preceding-text
…
Post-text …
ex:body-1 ex:target-1
ex:annotation-1
about
“Nursing mothers who are ultra-rapid metabolizers may also experience overdose symptoms such as extreme sleepiness, confusion, or shallow breathing.”
Biomedical Informatics62
Risk factors
patient characteristics X potential adverse event
patient characteristics X DDI mechanism
drug characteristics
route of administration, dose, timing, sequence
Biomedical Informatics63
Incidence
prevalence of co-prescription
prevalence of AE
incidence of AE in exposed and non-exposed
Biomedical Informatics64
Seriousness of the AE
Classified by specific clinical outcome
...but, can any seriousness ranking be generally accepted?
no effect death?
Biomedical Informatics65
Biomedical Informatics66
Linkage to DrugBank – Approach 1
1. FDA UNII table provides structure string:
2. NCI Resolver provides InChIKey:
3. DrugBank record with the above InChIKey provides identifier:
Results:429 out of 2,264 ingredients are linked, out of which 424 are
valid
“N7U69T4SZR”
Starting with UNII….
2-METHYL-4-(4-METHYL-1-PIPERAZINYL)-10H-THIENO(2,3-B)(1,5)BENZODIAZEPINE
KVWDHTXUZHCGIO-UHFFFAOYSA-N
DB00334
Idea: Using NCI Resolver & InChIKey
Biomedical Informatics67
Linkage to DrugBank – Approach 2
“OLANZAPINE”
1. ChEBI preferred name from NCBO Bioportal:
2. ChEBI identifier from NCBO Bioportal:
3. DrugBank record with the above ChEBI identifier provides identifier:
Results:718 out of 2,264 ingredients are linked, out of which 707 are
valid
“OLANZAPINE”
7735
DB00334
Idea: Using ChEBI identifier & NCBO Portal
Starting with name….
Biomedical Informatics68
Linkage to DrugBank – Approach 3
Starting with all data in the FDA UNII table and DrugBank….
1. Index all FDA UNII table and DrugBank XML attributes2. Search for linkage points and score similarity: UNII -> Substance Name DrugBank -> brands -> brand: 0.94 UNII -> Preferred Substance Name DrugBank -> name : 0.91 UNII -> Substance Name DrugBank -> synonyms -> synonym : 0.83 …3. Prune list of linkage points based on cardinality, coverage, and average score4. Establish links between FDA UNII table and DrugBank using the linkage points UNII “OLANZAPINE” DrugBank “Zyprexa” : 1.0 …Results: 1,179 out of 2,264 ingredients are linked, out of which 1,169 are valid
“N7U69T4SZR”
UNII
“OLANZAPINE”
Preferred Substance Name
“2-METHYL-4….”
Molecular Formula
“ZYPREXA”
synonym
Idea: Automatic discovery of linkage points
Biomedical Informatics69
Linkage Point Discovery Framework• A generic framework for unsupervised
discovery of linkage points
Details can be found at: O. Hassanzadeh et al. “Discovering Linkage Points over Web Data”. To Appear in PVLDB, Vol 6. Issue 6, August 2013