NTHU AI Lab 共頁,第 1 頁 Ontology-based Quantitative Simulation of Chemical Reactions in Biological Pathway Systems Biology Presentation
NTHU AI Lab 共頁,第 2 頁Outline System Architecture Web service for pathway database Knowledge presentation Quantify Chemical reactions Workflow
NTHU AI Lab 共頁,第 3 頁 System Architecture MetaCyc Pathway database Bio domain ontology KEGG Pathway database Wrapper agent Calculate agent Literature database Knowledge agent Triple transformation Knowledge extraction Simplify chemical networks Group the compounds calculates the coefficient Web service register Service description Information wrapper
NTHU AI Lab 共頁,第 4 頁 Web service for pathway database Using KEGG API to get the information, for example: Mark_all_pathway_by genes –User should input the name of the gene or the id of the gene, but it is hard to know the biological meaning for the machine –So we should develop the semantic meaning above the WSDL using OWL-S –Another, we should use the thesauri to know the type (genes, chemical compound, enzymes) of the input.
NTHU AI Lab 共頁,第 5 頁 KEGG pathway we parse the xml format of the pathway. After parsing the files, we transfer it to the triples. Finding the relationship between the two enzymes. Find the chemical reaction evolved in the pathway
NTHU AI Lab 共頁,第 6 頁 Transfer to the triples ( cpd:C00267, catalyze_with, ec: ) ( cpd:C00267, catalyze_with, ec: )
NTHU AI Lab 共頁,第 7 頁 Knowledge presentation Presents the concept of the pathways and reactions. The systems would know every compound and enzyme in those pathways. Biochemical reaction includes substrate, product, enzyme –substrate, product and enzyme are kinds of compound –enzyme catalyzes compound
NTHU AI Lab 共頁,第 8 頁 Quantify Chemical reactions Kinetic model Concentration & Reaction rate (Enzyme activity depends on concentration) Flux control & Elastic control & Deviation control Simplify chemical networks independent pathway group by branch points (Linear, Branched network)
NTHU AI Lab 共頁,第 9 頁Workflow If the user wants to increase the product of compound A for experiment, the systems would find the reactions involving compound A in pathway database such as KEGG, BioCyc…etc. Simplify chemical networks to independent network from triples Group the coefficients according to the branch points from triples Calculates the coefficient of each compound in biological pathway and some constraints about chemical reactions.
NTHU AI Lab 共頁,第 10 頁Pathway
NTHU AI Lab 共頁,第 11 頁 Goal & results Optimal experiment design In E. coli K-12 Pathway (mixed acid fermentation), E. Coli is suitable for biosynthesis of the ethanol. image?type=PATHWAY&object=FERMENTATION- PWY bin/get_pathway?org_name=eco&mapno= image?type=PATHWAY&object=FERMENTATION- PWY bin/get_pathway?org_name=eco&mapno=00010
NTHU AI Lab 共頁,第 12 頁Usage
NTHU AI Lab 共頁,第 13 頁 Examples (Ethanol) Pathway & reactions –Glycolysis (KEGG) Ethanol + NADP+ Acetaldehyde + NADPH –EC , EC Ethanol + NAD+ Acetaldehyde + NADH + H+ –EC , EC Ethanol + PQQ PQQH2 + Acetaldehyde –EC –Urethane + H2O Ethanol + CO2 + NH3 EC –Long-chain-fatty-acyl ethylester + H2O Long-chain fatty acid +Ethanol EC –trans-Cinnamoyl beta-D-glucoside + Ethanol Ethyl cinnamate +Glucose EC –Mercaptoethanol + Hydrogen cyanide Ethanol + Thiocyanate EC
NTHU AI Lab 共頁,第 14 頁 Examples (Ethanol) Pathway & reactions –Oxidative ethanol degradation II (MetaCyc) Ethanol+ NADPH + O2 H2O+NADP+acetaldehyde –Oxidative ethanol degradation III Ethanol+ H2O2 H2O+acetaldehyde –Oxidative ethanol degradation I…… Cetaldehyde+NADH ethanol+NAD –Carboxylic ester hydrolases H2O+a long-chain-acylethylester ethanol+afatty –In linear amides H2O+urethan NH3+CO2+ethanol
NTHU AI Lab 共頁,第 15 頁 Ontology-based Biological Pathway Modeling using Bayesian Network Systems Biology Presentation
NTHU AI Lab 共頁,第 16 頁Outline System Architecture Bayesian network
NTHU AI Lab 共頁,第 17 頁 System Architecture Pathway database Gene ontology Clustering Bayesian classification Bayesian Network … ……… Gene 1 Gene 2 Gene 3 Gene 4 Gene 5 Gene 6 Attribute independentCausal networkFunction Annotation
NTHU AI Lab 共頁,第 18 頁 Bayesian network They allow a subset of the variables conditionally independent. –A graphical model of causal relationships. Several cases of learning Bayesian belief networks: – Given both network structure and all the variables: easy. –Given network structure but only some variables. –When the network structure is not known in advance. Directed acyclic graph (DAG) + Conditional probability table (CPT)