CISBIC:Sub-project 1:

Stephen MuggletonCISBIC, Flowers Building, Imperial College London.www.imperial.ac.uk/cisbic

Modeling genotype-phenotype Modeling genotype-phenotype relations in Campylobacter: an relations in Campylobacter: an update and future plansupdate and future plans

Overview

•Machine Learning in Systems Biology

•CISBIC sub-project 1

•Hypotheses being tested

•Automated experiment selection

•Conclusions

Systems Biology: The CISBIC Vision

Machine Learning

BLPsPMDPsLogic Programs

SCFGsHMMsGrammars

Bayes’ netsNeural netsDecision trees

MixedProbabilisticLogical

Inductive Logic Programming

Background knowledge. Incomplete biological network.

Examples. Temporal traces of up/down regulation.

Hypothesis. Extra network annotation.

CISBIC subproject 1

Aim. Model effects of changes to pathogen genome of expression of glycosolated surface molecules involved in triggering of innate immune response of host.

Mycobacterium bovis Campylobacter jejuni

Initial focus

• Synthetic pathways for capsule in C. jejuni• 38 genes involved. Full set of knock-outs.• Functions - 1/3 known,1/3 suspected, 1/3 unknown. • Microarray and metabonomic experiments.

Polysaccharide capsule

Pathway database

Prolog database

Machine learning

Visualisation

KEGG BioCyc

ONDEXONDEX

Brendan/papers

Hypotheses for capsule pathway (in red)

cj1432c

cj1416c / cj1417c / cj1418c

Hypotheses predict

• Cj1416, Cj1417, Cj1418 involved in OMePN synthesis. Wren group has verified this by mutagenesis and structural analysis. Fairly obvious from BLAST analysis.

• Cj1432, a protein of unknown function, is central to capsule synthesis. Not predictable from amino acid similarity and BLAST analysis.

Experiments to test cj1432 hypothesis

1.Immunoassay.2.Alcian Blue dye staining of

Campylobacter.3.Electron microscopy – direct visual

inspection.4.Complementation – reintroduce cj1432

to chromosome of mutant5.Structural analysis of capsule glycan.6.Protein purification and enzyme assays.

Hypotheses/experiments

E1: A E2: B E3: C E4: D E5: E E6: AB E7: BC E8: BE E9: AD E10: ABC E11: ABCD E12: ABCEH1: codes(geneX,rA) 0 1 1 1 1 0 1 1 0 0 0 0H2: codes(geneX,rB) 1 0 1 1 1 0 0 0 1 0 0 0H3: codes(geneX,rC) 1 1 0 1 1 1 0 1 1 0 0 0H4: codes(geneX,rD) 1 1 1 0 1 1 1 1 0 1 0 1H5: codes(geneX,rE) 1 1 1 1 0 1 1 0 1 1 1 0

E

C B A

Pre-E

Pre-C

Pre-B

rB

rC

rD E

E10

E9 E11

H4H1

0

00

1

11

H5E8

H2 H3

Conclusions

•Integration of diverse background knowledge

•ILP produces readable rules

•Gap-filling in networks

•Ongoing NMR/MS experimental testing of hypotheses

• Cj1432 – potential C. jejuni vaccine