[PPT] - POLYTOMY REFINEMENT FOR THE CORRECTION OF DUBIOUS DUPLICATIONS IN PowerPoint Presentation

SLIDE 1

POLYTOMY REFINEMENT FOR THE CORRECTION OF DUBIOUS DUPLICATIONS IN GENE TREES

Manuel Lafond1, Cedric Chauve2,3, Riccardo Dondi4, Nadia El-Mabrouk1

1 Université de Montréal, Canada 2 Université Bordeaux 1, France 3 Simon Fraser University, Canada 4 Universitá degli Studi di Bergamo, Italy

SLIDE 2

Introduction

Gene tree for the SLC24a2 gene family (solute carrier 24)

G :

SLC24 Mouse SLC24 Rat SLC24 Microbat SLC24 Human SLC24 Chimp SLC24 Megabat SLC24 Squirrel

SLIDE 3

Introduction

Species tree for the species having a gene in G.

S :

Mouse Rat Microbat Human Chimp Megabat Squirrel

SLIDE 4

Introduction

G and S disagree

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl

G :

SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

SLIDE 5

Introduction

LCA MAPPING : associate each ancestral gene with the

species it belonged to

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl

G :

SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

u v w x y z v z w z z z

SLIDE 6

Introduction

G and S disagree => Duplication of an ancestral gene

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G :

SLC Mous

u v w x y z v z w z z z

SLIDE 7

Introduction

Extant species are expected to have 2 copies of the gene
None of them do. That’s dubious !

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : u v w x y z v z w z z z

SLIDE 8

Introduction

If some species was represented on both sides of the

duplication, it would be an Apparent Duplication (AD)

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : u v w x y z v z w z z z

SLC Hum

SLIDE 9

Introduction

Non-apparent duplication (NAD) : the left and right

subtrees of the duplication share no gene from the same species.

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : NAD u v w x y z v z w z z z

SLIDE 10

Introduction

Missing gene copies must have been lost sometime ago.
NADs usually imply a bunch of losses.

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : NAD

SLC Mous

u v w x y z v z w z z z

SLIDE 11

Introduction

NADs are called dubious, or ambiguous duplications in

the Ensembl database.

About 44% of duplication nodes are dubious.
The SLC24 gene tree has 32 duplication nodes, 24 of which are

dubious.

Simulations showed that only 5% percent of duplications

were actually NADs (Chauve & Mabrouk, 2009).

SLIDE 12

Introduction

Alternative scenario for the root of G : no duplication
ccurred.

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : NAD

SLC Mous

SLIDE 13

Introduction

S :

Mous Rat MicBat Hum Chmp MegBat Sqrl SLC MicBat/ MegBat SLC Hum/Mo/ Rat/Chmp/ Sqrl

NAD G :

Alternative scenario for the root of G : no duplication
ccurred => speciation => the bat genes should be

separated from the others.

SLIDE 14

Introduction

Break G as least as possible : send the maximal bat

subtrees left, and the maximal rodent/primate subtrees right

S :

Mo Rat MicBat Hum Chmp MegBat Sqrl SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G :

SLIDE 15

Introduction

SLC Mous SLC Rat SLC MicBat SLC Hum SLC Chmp SLC MegBat SLC Sqrl

G : G’ :

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl SLC MicBat SLC MegBat

Break G as least as possible : send the maximal bat

subtrees left, and the maximal rodent/primate subtrees right

SLIDE 16

Introduction

G’ ends up with possibly two unresolved polytomies.
We are looking for a binary refinement of these

polytomies.

G’ :

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl SLC MicBat SLC MegBat

SLIDE 17

Introduction

Other sources of polytomies :
Lack of phylogenetic signal in the sequences, causing some gene

tree construction algorithms to leave the gene tree partially unresolved.

Contraction of gene tree branches having low support (e.g.

bootstrap values).

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl

SLIDE 18

Previous works

Find a binary refinement minimizing:
Duplications + losses (Chang & Eulenstein, 2006, O(n3));
Duplications + losses (Lafond & Swenson & El-Mabrouk, 2012,

O(n))

Duplications and then losses (Zheng, Wu, Zhang, 2012, O(n))
Losses: It’s a linear problem.
Our problem here:

Minimize NAD nodes

For all these optimization criteria, polytomies can be

refined independantly. Thus we reduce the problem to a single polytomy.

SLIDE 19

Introduction

Given : a polytomy P and a species tree S
Find : a binary refinement of P that minimizes the number
f NADs created.

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl Rat Hum Chmp Sqrl Mous

P S

SLIDE 20

Introduction

Given : a polytomy P and a species tree S
Objective : find a binary refinement of P that minimizes

the number of NADs created.

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl Rat Hum Chmp Sqrl Mous SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl

NAD P S

SLIDE 21

Introduction

Given : a polytomy P and a species tree S
Objective : find a binary refinement of P that minimizes

the number of NADs created.

SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl Rat Hum Chmp Sqrl Mous SLC Mous SLC Rat SLC Hum SLC Chmp SLC Sqrl

P S

SLIDE 22