Data mining in genetics

Junior Barrera

BIOINFO-USP

Layout

• Introduction

• Data mining

• Mapping of rare genes

• Expression analysis: measure, genes differentially

expressed; clustering expression signals; identification of

gene regulation networks

Introduction

Knowledge evolution in genetics

• Heredity - Mendel (1866)

• The phenotypes of an individual depends on

genes of his parents.

Knowledge evolution in genetics

• Chromosome Theory - Morgan (1910)

• Genes were situated in chromosomes

Knowledge evolution in genetics

• The molecular structure of chromosomes

(Watson and Crick - 1953)

• DNA structure: the double helix

• Four basis: adenine(A), guanine(G),

thymine(T), cytosine(C)

• genes are sequences of nucleotides

Knowledge evolution in genetics

Knowledge evolution in genetics

• DNA manipulation

• cut, replication and decoding

Knowledge evolution in genetics

• Genetic engineering

• species modification, drug production

Knowledge evolution in genetics

• Genes control the metabolism

• Metabolism occurs by sequences of

enzyme-catalyzed reactions.

• Enzymes are specified by one or more

genes

Knowledge evolution in genetics

Knowledge evolution in genetics

• Gene expression

Data acquisition

Data acquisition

Data acquisition

Quantization - {-1,0,1}

Data mining

P1

P2

Pn

Pi : analytical and mining procedures (kernel parallel)

Objected oriented database

images Express

Table

Signal clusters

Dynamical

System

Sequence module

microarray module

Genbank

database

P1

P2

P3

Pn

P1

P2

P3

Pnquery

query

operational

operational

analitical

analitical

Access control module

Integrated Environment

query

Another

databases

Clinical data

Data Warehouse

M_database

Analysis tools writer SpreadSheet Dataminig

MOLAP/ROLAP server

IMS OthersRDMS

Wrapper Wrapper Wrapper

Integrator

Metadata Data

Mart

Users

1

2

3

4

System Architecture

DB/2DB/2

SybaseSybase

OracleOracle

InformixInformix

SliceSlice NNSliceSlice 11

Appl. n

Appl. 3

Appl. 2

Appl. 1

JavaJava

PBPB

DelphiDelphi

MATLAMATLA

N-2 slices

Consulte rules and libraries

Proteome

Transcriptome

Genome

Pathways

Σ

Wet LabWhat genes regulate the

pathway A->B->C->D ?

Knowledge

Medline

History of

discovering

Pi,Ni

P1

P2

Pi : analytical and mining procedures (kernel parallel)

Objected oriented database

Pn

N1

N2

Nn

Ni : knowledge discovering procedures (kernel parallel)

molecule

cell

tissue

organism

Protein structure

and dynamics

DNA, Protein,

Gene Expression,

Gene Networks

Population

E4000

E3500 V880

GRID Computer - DCC-IME-USP

Internet

and

Intranet

...

E4000:Databases and web application services

Processing services

(by SEFAZ)

10 processors

16 G-bytes main memory

1 T-bytes disk

(by CAGE-FAPESP)

4 processors

8 gigabytes

(by Malaria-FAPESP)

16 processors

32 G-bytes

V880?

(by eucalipto-CNPq ???)

16 processors

32 G-bytes

Mapping of rare genes

ACGAATCTAGAGAATTAATTAACCGAGTTAAGA

ACGAATCTAGAGAATTAATTAACCGAGTTAAGA

Exon

Training from known genes

Expression Analysis

Analysis Phases

MeasureClustering

G. Dif. Exp

Clustering

ClusteringG. R. Net.

Image Analysis

Selection of clones

1. Clusters of the same gene

5’ 3’Gene chosen

2. Choice of a representation for the gene

5’ 3’

Cuidado com sítios de Poliadenilação!!

Captura das Imagens Cy5 Cy3

Hibridization

cDNAs Misturas de cDNAsmarcados CY3 e CY5

Hirata R, Barrera J, Hashimoto R, Dantas D, Esteves G. In press, 2002.

Expression Calculus

Ratio

signal + noise

Forground

Background

Dispersion of cy3 and cy5 give an idea of the spot quality

Spot bom

Spot ruim

o Pixels do backgroundo Pixels do foreground

ScanAlyzeScanAlyze

Totalmente baseado em ROIs.

Amostragens de pixels simples

QuantArray®

SpotSpotSoftware todo automatizado

Jain, et al. Genome Research, 12: 325-332, 2002

Mas nem sempre funciona…

Então, correção manual!!!

Os pixels são escolhidos combase no histograma.

BIOINFO BIOINFO -- USPUSP

Example - Segmentation

Exemplo - Segmentação

cDNAs used

� Q gene - 637pb, 52,3% de C e G

� trpC - 338pb, 45,3% de C e G

� lysA - 303pb, 47,2% de C e G

� ST0280 (ORESTES) - 659pb, 34,6% de C e G

� IL6 - 948pb, 37,7% de C e G

� IRF1 - 2069pb, 52,5% de C e G

Dilution of fixed material

Each cDNA will be fixed in the dilutions 1/1, 1/2, 1/4, 1/8, 1/16

Legenda

B1

Diluição 1

Diluição 2

Diluição 3

Diluição 4

Diluição 5

B2

Legenda

Diluição 1

Diluição 2

Diluição 3

Diluição 4

Diluição 5

B3

Legenda

Diluição 1

Diluição 2

Diluição 3

Diluição 4

Diluição 5

B4

Legenda

Diluição 1

Diluição 2

Diluição 3

Diluição 4

Diluição 5

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B4B3

B2B1

For a good signal:

• the linear regression is a good estimator for

ratio (background estimation is avoided)

• Swap permits to normalize cy3 and cy5

• Confidence intervals increase inversely with

the signal intensity

Genes differentially expressed

Experiment

• Choose a population of men and measure

their physical characteristics

• Ask men to move a 150 kg object

• Separate men that succeed and the ones that

do not

• Find common characteristics between men

that succeed and between men that do not

Genes differentially expressed

• Choose a set of genes

• Measure the expression of these genes on

two different Biological states

• Choose subsets of genes that are enough to

characterize each Biological state

−3.5 −3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1−1.5

−1

−0.5

0

0.5

1

1.5

2

phosphofructokinase, platelet

v−

erb

−b2 a

via

n e

ryth

robla

stic leukem

ia v

iral oncogene h

om

olo

g 2

LINEAR CLASSIFIER (DISPERSED−GAUSSIAN) w/ σ = 0.600

1−th (0.389593)

7−th (0.416607)

BRCA1 BRCA2/sporadic

the tumor sample from Patient 20

Clustering

1 2 3 4 5 6 7-6

-4

-2

0

2

4

6

timelo

g2(r

atio)

time course data

C1

C2

C3

-40 -20 0 20 40 60 80 100 120 140-20

-10

0

10

20

30

40

50

60KL plot

C1

C2

C3

Dendrogram

Time course data

KL plotmultidimensional space

different views of different views of different views of

Microarray dataMicroarray dataMicroarray datamis-classified

Original clusters

Clustered by dendrogram

Example

No error!

Tighter clusters due

to small variance

Results from Fuzzy c-means

Gene Regulation Networks

Cellpeptide

other signals

mRNAGENES

NETWORK

Translat.Transcr.Proteins

Pool Pathways

Metabolic

Division Steps

Cell Cycle Modeling

u1

I

u2

u5

v

vfp

w1

w2

w1fp

w2fp

w1f

w2f

x1

x6

y1

y2

z

x1f

x3f

x4f

x6f

x6fp

x1fp

y1fp

y2fp

.

.

.

.

.

.

z

Forward SignalFeedback to pFeedback to previous layer

y1f

y1f

y2f

y2f

p

Modeling Dynamical Systems

Simulator Architecture

Knockout

u1

I

u2

u5

v

vfp

w1

w2

w1fp

w2fp

w1f

w2f

x1

x6

y1

y2

z

x1f

x3f

x4f

x6f

x6fp

x1fp

y1fp

y2fp

.

.

.

.

.

.

z

Forward SignalFeedback to pFeedback to previous layer

y1f

y1f

y2f

y2f

p

Division Steps

Challanges

• Architecture identification

• Dynamics transition function identification