Knowledge management for integrative omics data analysis
- 1. Knowledge Management for Integrative Omics Data Analysis Barcelona 15.02.13 Dr. Hilmar Ilgenfritz Biomax Informatics AG www.biomax.com
- 2. Biomax – Unique disruptive technology Biomax Profile Biomax Vision Headquartered near Munich Master scientific complexity Germany Reduce cost and time In business for more than 15 Ensure ease of use years Increase speed of development World wide customer base BioXM is a configurable Enable centers of excellence for knowledge management platform personalized medicine to flexibly interconnect isolated Support for Systems Biology silos of information in biomedical research
- 3. Generate scientific impact from knowledge exploitation Challenge Actionable to bridge knowledge for the gap rational decision making Insight with Aggregate scientific impact Collect
- 4. ... e.g. expression data in a pathway context
- 5. Why „Knowledge Management“? Knowledge: “the realisation and understanding of patterns and their implications existing in information” Need to mine information for patterns A pattern often only emerges when information from different silos is combined e.g. Expression with gene function, SNPs with clinical history of patients, ... Need semantically integrated information e.g. Information about identical or “equivalent” objects and “meaning” requires framework for integration methods to find “equivalent” “meaning”
- 6. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis • Knowledge extraction • Collaboration and project management
- 8. Knowledge aspects in Systems Medicine Ontologies/SOPs/Study Design Clinical Biobank Data eCRF/ Experimental EHR Data Public Literature KM Molecular Data Data Data integration
- 9. Multi-Scale Modelling Organs Cells Individuals Tissues requires “language” for formal, structured description >> semantic mapping
- 10. Semantic mapping Mapping entities or descriptions e.g. genes, phenotypes cancer - blastoma, model parameters etc. with “equivalent meaning” from different sources using controlled, structured vocabularies drop dead drd, Q8IR42, AAD52607
- 11. Traditional semantic mapping KEGG PubMed Gene Ontology UniProt
- 12. Working with semantic networks • Connected data, meta- data and knowledge • Query, view, report • Integrate with analysis
- 13. Machine is configured to deliver relevant actionable BioXM knowledge through apps Machine is configured to build the connections to information Knowledge and data, based on the Model knowledge model Any type of data Any format of data Any volume of data Any location of data Any size of data Documents Spread sheets
- 14. Apps Common users Knowledge Model Power user Information and Data Access Administrator
- 15. Concept - Agile Solution Building Step 1: Specification • Designing the data model Query the knowledge network, explore Define the domain-specific data the graph and report query results model Step 3: Use Step 2: • Query building Implementation and information • Importing Instantiate the retrieval information knowledge network with data and information from external resources
- 16. Solution deployment Step 4 Web Apps for Information Retrieval, Reporting and Annotation
- 17. Sketch a domain model describing the semantic network
- 18. Setting up the data model - graphical support for modeling semantic networks Example: sub-network defining a study design
- 19. Import data into semantic structure
- 20. Knowledge Network Representation Dynamic network representation in BioXM Each node or edge of the network may serve as entry point for further exploration!
- 21. Knowledge Network Expansion Dynamic network representation in BioXM
- 22. On-the-fly queries retrieve context-specific sub-network Experimental Knowledge data Molecular data Q u e r y e.g. Patient overview
- 23. Natural language query wizard
- 24. Explore network
- 25. Application areas Clinical Research Biobanking Pathway Analysis Literature Mining
- 26. Application areas NextGenSequencing Comparative Genomics Urban/rural ozone levels (annual average) Systems Biology Environmental Sciences
- 27. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis • Knowledge extraction • Collaboration and project management
- 28. Connecting different sources Example: Literature derived tissue specific FAS pathway + CTD
- 29. Connecting different sources Example: Literature derived tissue specific FAS pathway + CTD
- 30. Validation of interconnection CASP10 - Tetrachlorethene, evidence from CTD 3472561
- 31. Another example
- 32. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis • Knowledge extraction • Collaboration and project management
- 37. Navigate the network (find associated pathways)
- 38. Enter neighboring knowledge domain (Toll-like receptor signalling pathway)
- 39. Collect all information about IL6
- 40. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis Multi-variate data analysis • Knowledge extraction • Collaboration and project management
- 41. Complexity of chronic diseases Socio- Lifestyle-environment economic Risk and protective factors determinants Tobacco smoking, pollutants, allergens, nutrition, infections, physical exercise, others Gender Genes, Cells, Tissues, Organs Biological expression of chronic diseases Transcripts, proteins, metabolites, Target organ local inflammation, Systemic inflammation Age Cell and tissue remodeling Clinical expression of chronic diseases Co-morbidities, Severity of co-morbidities, Persistence remission, Long-term morbidity, Responsiveness - side effects to treatment
- 42. The unmet need – transform data into insight Insight in aggregated clinical data for patient stratification in chronic disease • Up to 6,000 parameters per patient • 5 years of patient history Disperse clinical records
- 43. Oversight and insight over the clinical processes
- 44. Patient map – parameters for outcome prediction Specific outcome group accumulates in certain areas of the map
- 45. Patient group differentiating outcome profile Out of almost XXX differentiating attributes at high confidence X attributes constitute a robust predictor
- 46. Reliable outcome prediction validation Patients with outcome risk can be predicted with high reliability.
- 47. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis Integrative analysis • Knowledge extraction • Collaboration and project management
- 48. COPD ROS hypothesis Muscle wasting in COPD patients is effect of systemic inflammation resulting in nitroso-redox imbalance by mitochondria respiratory chain uncoupling in COPD patients with low body mass index
- 49. Metabolism/ROS-production ODE model linked with clinical data (Selivanov, Cascante, Barcelona) Biophysical J. 92, 3492-3500 Glycolysis J. Theor. Biol. 252, 402-410 Bioinformatics 22, 2806-2812 Clinical Data connection NAD Glc BMC Neuroscience, 7,(Suppl 1):S7 Mitochondria O2 uptake Exhalates TCA cycle Cit from Biobridge NADH Pyr AcCoA NAD OAA NADH Succ ADP NAD Lac Omics in O Blood RESPIRATION 2 transportc La ROS, ATP glutathione etc… CrP antioxidant system Clinical Data connection ROS cell damage ”OMICS” in muscle biopsies from Biobridge (nitroso-redox balance, proteomics, genomics, signalling Inflamation markers …)
- 50. Probabilistic network connecting inflammation and metabolism baseds on omics data (Turan, Falciani, Birmingham) PLoS Comput Biol. 7 e1002129
- 51. Extending the deterministic model Glycolysis NAD Glc Clinical ADP Resulting connecting network data mechanic Myofibrils Glycolysis work ATP TCA cycle Cit NAD Glc NADH Pyr AcCoA NAD ADP OAA NADH Succ mechanic work ADP ATP O2 NAD Lac NADH Pyr transport Electron chain CrP ATP diffusion CrP ROS NAD Lac Deterministic models COPD knowledge base ATP Data clinical/ CrP experimental Selection of hubs Oxidative phosphorylation TCA COPD KB Cycle network Glycolysis search Probabilistic network Physiological measurments
- 52. Integrative prediction models ROS model gas exchange O IO O lung heterogeneities I I O IO O I I I I O O IO KM KM I I KM Simulation environment clinical data BioBridge PAC-COPD ECLIPSE
- 54. Semantic description of parameters
- 55. Mapping of model parameters Context:Parameter Description:Instance_B Ontology:A:54645 Element:Parameter:Instance_B Ontology:A:5461 Ontology:B:987723 Ontology:C:21365 Ontology:A:54632 Element:Compound:Oxygen Element:Model Parameter:Instance_A Context:Model Parameter Description:Instance_A
- 56. Development of new probabilistic Glycolysis NAD Glc network from COPD KB hetero ADP genic mechanic Myofibrils work ATP TCA cycle Cit NADH Pyr AcCoA Probabilistic NAD (predictive) OAA NADH Succ network ADP NAD Lac CrP Electron chain 0.1 0.3 ATP 0.4 0.6 O2 diffusion transport 0.8 CrP ROS 0.5 0.7 Deterministic models COPD knowledge base 0.2 0.4 0.1 Data clinical/ experimental Selection of hubs Oxidative phosphorylation TCA Cycle Resulting connecting network Glycolysis COPD KB Physiological Correlation network measurments network search
- 57. Hidden variable prediction ROS model gas exchange O I O O lung heterogeneities I I O I O O I I I I O O I O KM KM KM I I Simulation environment clinical data BioBridge PAC-COPD ECLIPSE Probabilistic model 0.1 0.3 0.4 0.6 0.8 0.5 0.7 0.2 0.4 0.1
- 58. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis Statistics examples • Knowledge extraction • Collaboration and project management
- 59. Principal Component Analysis of Functional Modules (activity of tissue remodelling pathways is altered in COPD patients)
- 62. Cluster analysis n = 92 n = 200
- 63. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis Network search • Knowledge extraction • Collaboration and project management
- 65. Network Search Results Resulting connectingonnecting network between two sources objects
- 66. Knowledge Management aspects • Data integration • Semantic mapping • Knowledge representation • Data analysis • Knowledge extraction • Collaboration and project management
- 68. Review results: Web Input form
- 69. Searches supporting the review flow
- 70. Results Stage 3: "include"
- 71. Thank you !