approaches for analyzing biological sequences

8/3/2019 Approaches for Analyzing Biological Sequences

1/33

New Approaches forNew Approaches for

Analyzing Biological SequencesAnalyzing Biological Sequences

Prof. Rushen Chahal Prof. Rushen Chahal

Prof. Rushen Chahal


2/33

ContributionsContributions

Developed methods for:

Identifying new genes

Constructing evolutionary trees

Comparing phylogenetic solution space

Prof. Rushen Chahal


3/33

B iological SequencesB iological Sequences

cgttaacaaagc... MAEKPKLH...

RNA pro teinD NA ( gene)

Prof. Rushen Chahal


4/33

M ain TasksM ain Tasks

F inding new genes:F inding new genes:Polymerase Chain Reaction (PCR)Polymerase Chain Reaction (PCR)CloningCloningGenomic sequencingGenomic sequencing

Determining gene functions:Determining gene functions:L ab work L ab work Related genes (homology)Related genes (homology)Genome comparisonGenome comparison

Prof. Rushen Chahal


5/33

Evolu tionary Tr ees

Evolu tionary Tr ees

time

Prof. Rushen Chahal


6/33

P utting it togetherP utting it together

?Ev olu tionEv olu tion

n ew seque nc es

database

related seque nc es

relatio n ships

Prof. Rushen Chahal


7/33

P rimer Selection forP rimer Selection forP olymerase Chain Reactions ( P CR)P olymerase Chain Reactions ( P CR)

G oal:G oal:

D iscover previously unknown genesD iscover previously unknown genesS trategy:S trategy:

D esign PCR primers for large set of knownD esign PCR primers for large set of known

gene family members gene family members

U nkn own gen es will (hopefully) be amplified U nkn own gen es will (hopefully) be amplified

Prof. Rushen Chahal


8/33

G ene FamilyG ene FamilyherpesEC

crnvHH2cmvHH3

humIL8bovLOR1RBS11 humSSR1

musdeltohumC5a

ratBK2 humTHRratRTA humMRG

humMAS

humfM LFratANGratG10d

dogRDC1

humRSC

chkGPCRgpPAF

musP2uchkP2y

ratODOR

ratLH bovOPhumEDG1

ratCGPCR

ratPOThumACTHhumMSH

musEP3humTXA2

musEP2ratCCKA

dogCCKB

dogAd1humD2

humA2ahamA1a

hamB2hum5HT1a

ratD1bovH1

humM1

ratNPYY1ratNK1flyNKflyNPY

musGIRratNTR

musTRHmusGnRHratV1a

bovETAmusGRP

Prof. Rushen Chahal


9/33

P olymerase Chain ReactionP olymerase Chain Reaction

Prof. Rushen Chahal


10/33

P rimersP rimers

C ommo n regio n => c ommo n primer

Prof. Rushen Chahal


11/33

P rimer G roupP rimer G roupherpesEC

crnvHH2cmvHH3

humIL8bovLOR1RBS11 humSSR1

musdeltohumC5a

ratBK2 humTHR

ratRTA humMRGhumMAS

humfML FratANGratG10d

dogRDC1

humRSC

chkGPCRgpPAF

musP2u

chkP2y

ratODOR

ratLH bovOP

humEDG1ratCGPCR

ratPOThumACTHhumMSH

musEP3humTXA2

musEP2

ratCCKA

dogCCKB

dogAd1humD2

humA2ahamA1a

hamB2hum5HT1a

ratD1bovH1

humM1

ratNPYY1ratNK1

flyNKflyNPY

musGIRratNTR

musTRHmusGnRHratV1a

bovETAmusGRP

?

?

?

Prof. Rushen Chahal


12/33

P rimer Selection P roblemP rimer Selection P roblem

O ptimal Primer S election Problem:O ptimal Primer S election Problem:

input:input: set of DNA sequencesset of DNA sequencesoutput:output: optimal set of primersoptimal set of primers

Theorem: NPTheorem: NP- -completecompleteProof: reduction from set coverProof: reduction from set cover

Prof. Rushen Chahal


13/33

A pproachesA pproaches

Ex act algorithms:Ex act algorithms:ex haustive bruteex haustive brute- -forceforcebranchbranch- -andand--boundbound

ProvablyProvably- -good heuristics:good heuristics:solution quality:solution quality: log(# sequences)log(# sequences) OPT OPT

Prof. Rushen Chahal


14/33

Ex tension: Ine x act P rimersEx tension: Ine x act P rimers

Goal: Optimize mismat c hes & #primers

Prof. Rushen Chahal


15/33


16/33

The B ig P ictureThe B ig P icture

?Ev olu tionEv olu tion

Identifying new genes using PCR

Prof. Rushen Chahal


17/33

Ev olutionary Tree ReconstructionEv olutionary Tree Reconstruction

N P-complete [Foulds & Graham 1982, D ay 1987]

Op timality Criteria:L east-Sq uar esMinim um-Ev olu tionMaxim um-Pa r sim onyMaxim um-L ike lihood

tree c ost

Prof. Rushen Chahal


18/33

P re v ious A pproachesP re v ious A pproaches

F itchF itch--Margoliash [1967]Margoliash [1967]NeighborNeighbor- -Joining [1987]Joining [1987]Quartet Quartet- -Puzzling [1997]Puzzling [1997]S plit S plit--Decomposition [1995]Decomposition [1995]PAUP [1998], PHYLIP [1993]PAUP [1998], PHYLIP [1993]

All use greedy & target best solutio nProf. Rushen Chahal


19/33

H owe v er .H owe v er .

Topologically distant solutions may e x ist Topologically distant solutions may e x ist

Dete c t diverse low c ost solutio n s

1

45 2

3

6

0.25620.2562

1

23 4

5

6

0.25600.2560

34

1

0.25610.2561

33

33

33

Prof. Rushen Chahal


20/33

[M addiso n 1991, Pe nn y 1995, Swofford 1997]

Random Starting Trees + H euristics?Random Starting Trees + H euristics?

Prof. Rushen Chahal


21/33

N eighborN eighbor- -Joining M ethodJoining M ethod

1

2

3

4

5

3

4

51

2

1

2

1

2

1

2

1

2

1

2

3

4

51

2 4

5

4

5

4

5

4

5

4

5

3

1

2

Prof. Rushen Chahal


22/33

G eneralized N eighborG eneralized N eighbor- -JoiningJoining1

2 3

4

5

3

1

2

4

51

43

2

53

42

1

5

1

2 4

5

3

3

4 1

5

2

3

4 2

5

1Prof. Rushen Chahal


23/33

G eneralized N eighborG eneralized N eighbor- -JoiningJoining

Controlling solution space sampling:Controlling solution space sampling:

K : max # partial solutions maintained K : max # partial solutions maintained Q (quality): # candidates selected for low cost Q (quality): # candidates selected for low cost D (diversity): # candidates selected for variety D (diversity): # candidates selected for variety

Tradeoff quality & topological diversity:Tradeoff quality & topological diversity:K = Q + DK = Q + D

Prof. Rushen Chahal


24/33

G N J P erformance (8 lea v es)G N J P erformance (8 lea v es)

least-squares cost least-squares cost

exhaustive

50, 045, 525, 255, 45

0, 50

Q D

1

1 0

1 00

0

1

2

3

4

5

6

n u m

b e r o

f s o

l u t i o n s

t o p o

l o g

i c a l

d i s t a n c e

0.00 1 0.01 0.1 1 0.001 0.01 0.1 1

Prof. Rushen Chahal


25/33

Solution Cost (16 lea v es)Solution Cost (16 lea v es)

meth ods

s o l u t i o n c o s t

LS

ME

1 0 -2

1 0 -3

10

-4

1 0 -5

1 0 -6

1 0 -7

1 0 -8

K=1 K=20 K=100

Prof. Rushen Chahal


26/33

Solution Di v ersity (16 lea v es)Solution Di v ersity (16 lea v es)

LS -maxME -maxLS -aveME -ave

0

2

4

6

8

1 0

1 2

t o p o

l o g

i c a l

d i s t a n c e

meth ods

K=1 K=20 K=100

Prof. Rushen Chahal


27/33

G N J Time Comple x ityG N J Time Comple x ity

G enerate candidates:G enerate candidates:O( K N O( K N 2 2 ) )

S elect candidates:S elect candidates:O( K N O( K N 2 2 (lg K + lg N )) (lg K + lg N ))

O(K N 3 (lg K + lg N)) G NJ ru n times (sec onds)

K N =8 N =16 N =32

20 0 .08 0 .8 9 .8

50 0 .2 2 .1 25 .1

1 00 0 .5 4 .4 52 .1

200 1 .1 8 .8 1 0 3 .7

500 3 .1 24 .2 262 .7

Prof. Rushen Chahal


28/33

SummarySummary

Ev olu tionEv olu tion

Identifying new genes using PCR

Detecting low-cost diverse trees

Prof. Rushen Chahal


29/33

Refereed P ublicationsRefereed P ublications

Pearson, W. R., Robins, G ., and Zhang, T., G eneralized NeighborPearson, W. R., Robins, G ., and Zhang, T., G eneralized Neighbor- -Joining: More Reliable Phylogenetic Tree Reconstruction, toJoining: More Reliable Phylogenetic Tree Reconstruction, toappear in Journal of Molecular Biology and E volution.appear in Journal of Molecular Biology and E volution.

Pearson, W. R., Robins, G ., Wrege, D. E ., and Zhang, T., O n thePearson, W. R., Robins, G ., Wrege, D. E ., and Zhang, T., O n thePrimer S election Problem for Polymerase Chain ReactionPrimer S election Problem for Polymerase Chain ReactionEx periments, Discrete and Applied Mathematics, Vol. 71, 1996, pp.Ex periments, Discrete and Applied Mathematics, Vol. 71, 1996, pp.231231--246.246.

Pearson, W. R., Robins, G ., Wrege, D. E ., and Zhang, T., A NewPearson, W. R., Robins, G ., Wrege, D. E ., and Zhang, T., A NewApproach to Primer S election in Polymerase Chain ReactionApproach to Primer S election in Polymerase Chain ReactionEx periments, Proc. International Conference on Intelligent Ex periments, Proc. International Conference on Intelligent S ystems for Molecular Biology, Cambridge, E ngland, July, 1995, pp.S ystems for Molecular Biology, Cambridge, E ngland, July, 1995, pp.

285285--291.291.Prof. Rushen Chahal


30/33

Refereed P ublications (cont.)Refereed P ublications (cont.)

G riffith, J., Robins, G ., S alowe, J. S ., and Zhang, T., Closing the G ap:G riffith, J., Robins, G ., S alowe, J. S ., and Zhang, T., Closing the G ap:NearNear--O ptimal S teiner Trees in Polynomial Time, I EEE TransactionsO ptimal S teiner Trees in Polynomial Time, I EEE Transactionson Computeron Computer- -Aided Design of Integrated Circuits and S ystems, Vol.Aided Design of Integrated Circuits and S ystems, Vol.13, No. 11, November 1994, pp. 135113, No. 11, November 1994, pp. 1351- -1365.1365.

Barrera, T., G riffith, J., McKee, S . A., Robins, G ., and Zhang, T., TowardBarrera, T., G riffith, J., McKee, S . A., Robins, G ., and Zhang, T., Towarda S teiner E ngine: E nhanced S erial and Parallel Implementations of a S teiner E ngine: E nhanced S erial and Parallel Implementations of the Iterated 1the Iterated 1- -S teiner MR S T Algorithm, Proc. G reat Lakes S ymposiumS teiner MR S T Algorithm, Proc. G reat Lakes S ymposium

on VLS I, Kalamazoo, MI, March 1993, pp. 90on VLS I, Kalamazoo, MI, March 1993, pp. 90- -94.94.

Barrera, T., G riffith, J., Robins, G ., and Zhang, T., Narrowing the G ap:Barrera, T., G riffith, J., Robins, G ., and Zhang, T., Narrowing the G ap:NearNear--O ptimal S teiner Trees in Polynomial Time, Proc. I EEE O ptimal S teiner Trees in Polynomial Time, Proc. I EEE International A S IC Conference, Rochester, S eptember 1993, pp. 87International A S IC Conference, Rochester, S eptember 1993, pp. 87- -90.90. Prof. Rushen Chahal


31/33

G eneralizationG eneralization

G ene r ate Par tial Solu tions

Ev aluate Par tial Solu tions

Select Par tial Solu tions

n- i & i

Parsimo n y, Least -Squares

Prefer dista n t trees

Prof. Rushen Chahal


32/33

Future WorkFuture Work

G eneralize G NJG eneralize G NJother optimality criteriaother optimality criteria

other solution space samplingother solution space samplingalternative topological distance metricsalternative topological distance metrics

Examine solution space using

GNJ

Examine solution space using

GNJ

Identify pathological data setsIdentify pathological data sets

Prof. Rushen Chahal


33/33

G eneralized N eighborG eneralized N eighbor- -JoiningJoining

1 . T = { t}, wher e t is the star -tr ee ov er S2 . R e peatT* n All next-ste p tr ees der iv ed fro m TT n Select up to K tr ees from T*

Unti l (all tr ees in T ar e fully r esolv ed)3 . Ou t pu t T

In pu t: a set of l eav es S,the distance mat r ix ov er SOu t pu t: a set of po ssib le phylo genetic tr ees for S

Prof. Rushen Chahal

approaches for analyzing biological sequences

Documents