Functional Linkages between Proteins. Introduction Piles of Information Flakes of Knowledge AGCATCCGACTAGCATCAGCTAGCAGCAGA CTCACGATGTGACTGCATGCGTCATTATCTA.

Functional Linkages between Proteins

Introduction

Piles of Information Flakes of KnowledgeAGCATCCGACTAGCATCAGCTAGCAG

CAGACTCACGATGTGACTGCATGCGTCATTATCTAGTATGAAAAAAGCCATGCTAGGCTAGTCAGCGACATGAGCCATGACTAGCGCAGCATCAGTCATCAGTCAGCGGAGCGAGGAGAGAGAGACGACTGACTAGCATGCACACATGCATGACGTCATGACTGCATGACTGACTGACTGACTGCATGCATGATATTTTTTTTTTCATGCATGCAGCATGCTACCCAGCTACAGTGCACAGCAGGTACGACGCATCAGCATACGTACGGCATGACGACTCAGACTACGCATACGACTACGAC

E. Coli S. cerevisiaeDroso

phila

Data Analysis

Traditional Methods (Experiments & Sequence Homology) The function of a protein

New Computational MethodsFunctional linkages between proteins

What does Functional Linkage mean ?

1) A common structural complex

2) A common metabolic pathway

3) A common biological process

4) All answers are correct

New Computational Methods

Phylogenetic Profile Method Rosetta Stone Method Chromosomal Proximity Method COG Database

11

Phylogenetic Profile Method

Phylogenetic Profile Method

Biologically: Simliar profile likelihood for common pathway or complex

Mathematically: N genomes 2N possible profiles A unique characterization

Why Should it Work ?

Rosetta Stone Method

Rosetta Stone Method (= Domain Fusion Analysis) Interacting proteins have

homologs in another organism fused into a single protein chain

Rosestta Stone Method

Rosestta Stone Method

Experimentally: E. coli ~4300 proteins ~6800 pairs similar to a single protein

Biologically:

Why Should it Work ?

Rosestta Stone Method

Validation Tests(E. coli):1) Annotation of proteins from the

SWISS-PROT database (68% vs. 15%)

2) Database of Interacting Proteins (6.4%)

3) Phylogenetic Profile Method (5% vs. 0.6%)

Models’ Success & Failure

+ -+ True

positiveFalse negative

- False positive

True negative

predicted

found

Rosestta Stone Method

False Negatives1) interactions that have evolved

through other mechanisms, i.e. there never was a fusion

2) The fused protein has disppeared during evolution

Rosestta Stone Method

False Positives1) Proteins have been fused to

regulate co-expression2) Can’t distinguish between binding

and non-binding homologs.3) Functional interaction rather than

a physical interaction

Rosestta Stone Method

Reducing Errors

Rosestta Stone Method

Reconstruction of metabolic pathways

Functional Protein Networks

Orthologs vs. Paralogs

Orthologs: genes in different species that evolved from a common ancestral gene by speciation

Paralogs: genes related by duplication within a genome

Chromosomal Proximity

Proximate Genes On the same strand Within 300 bp, or - Respective paralogs within 300 bp

Inferred link genes whose orthologs are close in

at least three phylogenetic groups

Chromosomal Proximity

Direct Link two proximate genes that are also

proximate in at least two other phylogenetic groups

Indirect Linkgenes whose orthologs are close in at least three other phylogenetic groups

Chromosomal Proximity

Chromosomal Proximity

Biologically: Conservation of proximity across multiple genomes Linked function

Logically: How likely is it that two genes are randomly proximate ?

Why Should it Work ?

Chromosomal Proximity

Method’s Reliability:

Chromosomal Proximity

1586 links were detected between ortholog families

KEGG: 80% in the same biological pathway

COG: 67% in the same functional category

Validation:

Chromosomal Proximity

Total validated links per genome

380 direct 352 inferred

Chromosomal Proximity

The COG Database

Clusters of Orthologous Groups COGs creation Each COG contains proteins that

have evolved from an ancestral protein

The COG Database

Current Numbers (2004) 43 Complete genomes 30 phylogenetic groups 2223 phylogenetic patterns 17 functional categories 3307 COGS 74059 proteins, 71% of total

The COG Database

The COG Database

Direct Information Annotation of Proteins

(group and individual) Phylogenetic Patterns Multiple Alignment

How can we use it ?

The COG Database

Detecting Missed Genes Patterns that contain all but one Mostly small proteins

How can we use it ?

The COG Database

Groups number growth

Are we approaching saturation ?

COG on the WWW

Reliability of the Methods

Major validation: Experimentally known linkages

Validation by “keyword recovery” search

references1) Eisenberg D, Marcotte EM, Xenarios I, Yeates TO. Protein function in

the post-genomic era. Nature. 2000 405:823-826. Review2) Marcotte EM, Pellegrini M, Ng HL, Rice DW, Yeates TO, Eisenberg D.

Detecting protein function and proteing protein interactions from genome sequences. Science. 1999 285:751-753.

3) Yanai I, Mellor JC, DeLisi C. Identifying functional links between genes using conserved chromosomal proximity. Trends Genet. 2002 18:176-179.

4) Tatusov RL, Natale DA, Garkavtsev IV, Tatusova TA, Shankavaram UT, Rao BS, Kiryutin B, Galperin MY, Fedorove ND, Koonin EV. The COG database: new developments in phylogenetic classification of proteins from complete genomes. Nucleic Acids Res. 2001 29:22-28.

5) Tatusov,R.L., Koonin,E.V. and Lipman,D.J. (1997) A genomic perspective on protein families. Science, 278, 631–637.

6) http://www.ncbi.nlm.nih.gov/COG