Lecture 14 Full Text - ut text indexing Jaak Vilo ... A suffix tree T for a string S ... • Each...

TextAlgorithms(6EAP)

Fulltextindexing

JaakVilo2016fall

1MTAT.03.190TextAlgorithmsJaakVilo

Problem

• GivenPandS– findallexactorapproximateoccurrencesofPinS

• YouareallowedtopreprocessS(andP,ofcourse)

• Goal:tospeedupthesearches

E.g.Dictionaryproblem

• DoesPbelongtoadictionaryD={d1,…,dn}– BuildabinarysearchtreeofD– B-TreeofD– Hashing– Sorting+Binarysearch

• Buildakeywordtrie:searchinO(|P|)– Assumingalphabethasuptoaconstantsizec– SeeAho-Corasickalgorithm,Trieconstruction

Sortedarrayandbinarysearch

he

hershis

global

indexhappy

head

header

info

informal

search

show

stop

1 13

Sortedarrayandbinarysearch

he

hershis

global

indexhappy

head

header

info

informal

search

show

stop

1 13

O( |P| log n )

TrieforD={he,hers,his,she}

0

1

2

h

e

3

s

4

5

e

h

8

i

7

s

9

r

6

s

O( |P| )

S!=setofwords

• Soflengthn

• Howtoindex?

• Indexfromeverypositionofatext

• Prefixofeverypossiblesuffixisimportant

a

b

b

aaa

aa

b

b

b

babaababaabbaabaababb

Trie(babaab)

b

a

a

b

Suffixtree• Definition: Acompactrepresentationofatriecorrespondingtothe

suffixesofagivenstringwhereallnodeswithonechildaremergedwiththeirparents.

• Definition(suffixtree).AsuffixtreeTforastringS(withn=|S|)isarooted,labeledtreewithaleafforeachnon-emptysuffixofS.Furthermore,asuffixtreesatisfiesthefollowingproperties:

• Eachinternalnode,otherthantheroot,hasatleasttwochildren;• Eachedgeleavingaparticularnodeislabeledwithanon-emptysubstring

ofSofwhichthefirstsymbolisuniqueamongallfirstsymbolsoftheedgelabelsoftheedgesleavingthisparticularnode;

• Foranyleafinthetree,theconcatenationoftheedgelabelsonthepathfromtheroottothisleafexactlyspellsoutanon-emptysuffixofs.

• DanGusfield: AlgorithmsonStrings,Trees,andSequences:ComputerScienceandComputationalBiology.Hardcover- 534pages1stedition(January15,1997).CambridgeUnivPr(Short);ISBN:0521585198.

Literatureonsuffixtrees• http://en.wikipedia.org/wiki/Suffix_tree• DanGusfield: AlgorithmsonStrings,Trees,andSequences:Computer

ScienceandComputationalBiology.Hardcover- 534pages1stedition(January15,1997).CambridgeUnivPr(Short);ISBN:0521585198.(pages:89--208)

• E.Ukkonen.On-lineconstructionofsuffixtrees.Algorithmica,14:249-60,1995. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.10.751

• Ching-FungCheung,JeffreyXuYu,HongjunLu."ConstructingSuffixTreeforGigabyteSequenceswithMegabyteMemory,"IEEETransactionsonKnowledgeandDataEngineering,vol.17,no.1,pp.90-105,January,2005.http://www2.computer.org/portal/web/csdl/doi/10.1109/TKDE.2005.3

• CPMarticlesarchive:http://www.cs.ucr.edu/~stelo/cpm/

• MarkNelson.FastStringSearchingWithSuffixTreesDr.Dobb'sJournal,August,1996.http://www.dogma.net/markn/articles/suffixt/suffixt.htm

http://stackoverflow.com/questions/9452701/ukkonens-suffix-tree-algorithm-in-plain-english

12

ThesuffixtreeTree(T)ofT

• datastructuresuffixtree, Tree(T),iscompactedtrie thatrepresentsallthesuffixesofstringT

• linearsize:|Tree(T)|=O(|T|)• canbeconstructedinlineartimeO(|T|)• hasmyriadvirtues (A.Apostolico)• iswell-known:366000Googlehits

E. Ukkonen: http://www.cs.helsinki.fi/u/ukkonen/Erice2005.ppt

13

Suffix tree andsuffix array techniques forpatternanalysis instringsEskoUkkonenUniv Helsinki

Erice School30Oct 2005E. Ukkonen: http://www.cs.helsinki.fi/u/ukkonen/Erice2005.ppt

Partlybasedon:

High-throughput genome-scale sequence analysis andmapping using compressed datastructures

VeliMäkinenDepartmentofComputerScience

University ofHelsinki

14

ttttttttttttttgagacggagtctcgctctgtcgcccaggctggagtgcagtggcgggatctcggctcactgcaagctccgcctcccgggttcacgccattctcctgcctcagcctcccaagtagctgggactacaggcgcccgccactacgcccggctaattttttgtatttttagtagagacggggtttcaccgttttagccgggatggtctcgatctcctgacctcgtgatccgcccgcctcggcctcccaaagtgctgggattacaggcgt


15

Analysisofastringofsymbols

• T=hattivatti’text’• P=att’pattern’

• FindtheoccurrencesofPinT:hattivatti

• Patternsynthesis:#(t)=4#(atti)=2#(t****t)=2


ISMB2009Tutorial VeliMäkinen:"...analysisandmapping..." 16

Solution:backtrackingwithsuffixtree

...ACACATTATCACAGGCATCGGCATTAGCGATCGAGTCG.....

17

Patternfinding&synthesisproblems• T=t1t2 …tn,P=p1p2 …pn ,stringsofsymbolsinfinitealphabet

• Indexingproblem:PreprocessT(buildanindexstructure)suchthattheoccurrencesofdifferentpatternsPcanbefoundfast– statictext,anygivenpatternP

• Patternsynthesisproblem:LearnfromTnewpatternsthatoccursurprisinglyoften

• Whatisapattern?Exactsubstring,approximatesubstring,withgeneralizedsymbols,withgaps,…

18

1. Suffix tree

2. Suffix array

3. Some applications

4. Finding motifs


19

ThesuffixtreeTree(T)ofT

• datastructuresuffixtree, Tree(T),iscompactedtrie thatrepresentsallthesuffixesofstringT

• linearsize:|Tree(T)|=O(|T|)• canbeconstructedinlineartimeO(|T|)• hasmyriadvirtues (A.Apostolico)• iswell-known:366000Googlehits


20

Suffixtrieandsuffixtree

a

b

b

aaa

aa

b

b

b

abaabbaabaababb

Trie(abaab)


21

Suffixtrieandsuffixtree

a

b

b

aaa

aa

b

b

b

a

baab

baabab

abaabbaabaababb

Trie(abaab) Tree(abaab)

22

Trie(T)canbelarge

• |Trie(T)|=O(|T|2)• badexample:T=anbn

• Trie(T)canbeseenasaDFA:languageaccepted=thesuffixesofT

• minimizetheDFA=>directedcyclicwordgraph(’DAWG’)

23

Tree(T)isoflinearsize

• onlytheinternalbranchingnodesandtheleavesrepresentedexplicitly

• edgeslabeledbysubstringsofT• v=node(α)ifthepathfromroottovspellsα• one-to-onecorrespondenceofleavesandsuffixes

• |T|leaves,hence<|T|internalnodes• |Tree(T)|=O(|T|+size(edgelabels))

24

Tree(hattivatti)hattivatti

attivattittivatti

tivattiivatti

vattiatti

ttiti

i

hattivattiattivatti ttivatti

tivatti

ivatti

vatti

vattivatti

attiti

i

i

tti

ti

t

i

vatti

vatti

vatti

hattivatti

atti

25


attivattittivatti

tivattiivatti

vattiatti

ttiti

i


tivatti

ivatti

vatti

vattivatti

attiti

i

i

tti

ti

t

i

vatti

vatti

vatti

hattivatti

hattivatti

atti

substring labels of edges represented as pairs of pointers

26


attivattittivatti

tivattiivatti

vattiatti

ttiti

i1 2 3

4

5

6

6,106,10

2,54,5

i

10

8

9

3,3

i

vatti

vatti

vatti

hattivatti

hattivatti

7

27

Tree(T)isfull textindexTree(T)

P

31 8

P occurs in T at locations 8, 31, …

P occurs in T ó P is a prefix of some suffix of T ó Path for P exists in Tree(T)

All occurrences of P in time O(|P| + #occ)

28

Findatt fromTree(hattivatti)hattivatti

attivattittivatti

tivattiivatti

vattiatti

ttiti

i


tivatti

ivatti

vatti

vattivatti

attiti

2

i

tti

ti

t

i

vatti

vatti

vatti

hattivatti

atti7

29

LineartimeconstructionofTree(T)

hattivatti

attivattittivatti

tivattiivatti

vattiatti

ttiti

i

Weiner (1973),

’algorithm of the year’

McCreight (1976)

’on-line’ algorithm (Ukkonen 1992)

30

On-lineconstructionofTrie(T)

• T=t1t2 …tn$• Pi =t1t2 …ti i:th prefix ofT• on-lineidea:updateTrie(Pi) toTrie(Pi+1)• =>verysimpleconstruction

31

Trie(abaab)

a a

b

b a

b

b

aa

Trie(a) Trie(ab) Trie(aba)

chain of links connects the end points of current suffixes

abaabaaaaεaε

32

Trie(abaab)

a a

b

b a

b

b

aa

a

b

b

aaa

aa

Trie(abaa)

33

Trie(abaab)

a a

b

b a

b

b

aa

a

b

b

aaa

aa

Trie(abaa)

Add next symbol = b

34

Trie(abaab)

a a

b

b a

b

b

aa

a

b

b

aaa

aa

Trie(abaa)

Add next symbol = b

From here on b-arc already exists

35

Trie(abaab)

a a

b

b a

b

b

aa

a

b

b

aaa

aa

a

b

b

aaa

aa

b

b

b

Trie(abaab)

36

WhathappensinTrie(Pi) =>Trie(Pi+1) ?

ai

ai

ai

ai

ai

ai

Before

After

New nodes

New suffix links

From here on the ai-arc exists already => stop updating here

37

WhathappensinTrie(Pi) =>Trie(Pi+1) ?

• time:O(sizeofTrie(T))• suffixlinks:

slink(node(aα))=node(α)

38

On-lineprocedureforsuffixtrie

1. Create Trie(t1): nodes root and v, an arc son(root, t1) = v, and suffix links slink(v) := root and slink(root) := root

2. for i := 2 to n do begin3. vi-1 := leaf of Trie(t1…ti-1) for string t1…ti-1 (i.e., the deepest leaf)

4. v := vi-1; v´ := 0

5. while node v has no outgoing arc for ti do begin

6. Create a new node v´´ and an arc son(v,ti) = v´´

7. if v´ ≠ 0 then slink(v) := v´´

8. v := slink(v); v´ := v´´ end9. for the node v´´ such that v´´= son(v,ti) do

if v´´ = v´ then slink(v’) := root else slink(v´) := v´´

39

Suffixtreeson-line

• ’compactedversion’oftheon-linetrieconstruction:simulatetheconstructiononthelinearsizetreeinsteadofthetrie=>timeO(|T|)

• alltrienodesareconceptuallystillneeded=>implicit andreal nodes

40

Implicitandrealnodes

• Pair(v,α)isanimplicitnode inTree(T)ifvisanodeofTreeandα isa(proper)prefixofthelabelofsomearcfrom v.Ifα istheemptystringthen (v,α)isa ’real’ node(=v).

• Let v=node(α´)in Tree(T). Then implicitnode(v,α)representsnode(α´α)ofTrie(T)

41

Implicitnode

…

v

(v, α)α…

α´

42

Suffixlinksandopenarcs

…

v

aα

…

α

root

slink(v)

label [i,*] instead of [i,j] if w is a leaf and j is the scanned position of T

w

43

Bigpicture

… … …

…

…

suffix link path traversed: total work O(n)

new arcs and nodes created: total work O(size(Tree(T))

44

On-lineprocedureforsuffixtree

Input: string T = t1t2 … tn$

Output: Tree(T)

Notation: son(v,α) = w iff there is an arc from v to w with label α

son(v,ε) = v

Function Canonize(v, α):

while son(v, α´) ≠ 0 where α = α´ α´´, | α´| > 0 do

v := son(v, α´); α := α´´

return (v, α)

45

Suffix-treeon-line:mainprocedure

Create Tree(t1); slink(root) := root(v, α) := (root, ε) /* (v, α) is the start node */for i := 2 to n+1 do

v´ := 0while there is no arc from v with label prefix αti do

if α ≠ ε then /* divide the arc w = son(v, αη) into two */son(v, α) := v´´; son(v´´,ti) := v´´´; son(v´´,η) := w

elseson(v,ti) := v´´´; v´´ := v

if v´ ≠ 0 then slink(v´) := v´´v´ := v´´; v := slink(v); (v, α) := Canonize(v, α)

if v´ ≠ 0 then slink(v´) := v(v, α) := Canonize(v, αti) /* (v, α) = start node of the next round */

http://stackoverflow.com/questions/9452701/ukkonens-suffix-tree-algorithm-in-plain-english

47

Theactualtimeandspace• |Tree(T)|isabout20|T|inpractice• brute-forceconstructionisO(|T|log|T|)forrandomstringsastheaveragedepthofinternalnodesisO(log|T|)

• differencebetweenlinearandbrute-forceconstructionsnotnecessarilylarge(Giegerich&Kurtz)

• truncatedsuffixtrees:ksymbolslongprefixofeachsuffixrepresented(Naetal.2003)

• alphabetindependentlineartime(Farach1997)

abcabxabcd

ApplicationsofSuffixTrees• DanGusfield: AlgorithmsonStrings,Trees,andSequences:

ComputerScienceandComputationalBiology.Hardcover-534pages1stedition(January15,1997).CambridgeUnivPr(Short);ISBN:0521585198.- book

• APL1:ExactStringMatchingSearchforPfromtextS.Solution1:buildSTree(S)- oneachievesthesameO(n+m)asKnuth-Morris-Pratt,forexample!

• SearchfromthesuffixtreeisO(|P|)• APL2:ExactsetmatchingSearchforasetofpatternsP


C

Backtobacktracking

AC

T

4 2 1 5 36

CT

TAT

TAC

T

AT C

T

A

ACA, 1 mismatch

Same idea can be used to many otherforms of approximate search, like Smith-Waterman, position-restricted scoringmatrices, regular expression search, etc.

ApplicationsofSuffixTrees

• APL3:substringproblemforadatabaseofpatternsGivenasetofstringsS=S1,...,Sn--- adatabaseFindallSithathavePasasubstring

• GeneralizedsuffixtreecontainsallsuffixesofallSi• QueryintimeO(|P|),andcanidentifytheLONGESTcommonprefixofPinallSi


• APL4:Longestcommonsubstringoftwostrings• FindthelongestcommonsubstringofSandT.• OveralltherearepotentiallyO(n2 )suchsubstrings,ifnisthelengthofashorterofSandT

• Donald Knuthonce(1970)conjectured thatlinear-timealgorithmisimpossible.

• Solution:constructtheSTree(S+T)andfindthenodedeepestinthetreethathassuffixesfrombothSandTinsubtreeleaves.

• Ex:S=superiorcalifornialives T=sealiver havebothasubstringalive.


Simpleanalysistask:LCSS

• LetLCSSA(A,B) denotethelongestcommonsubstringtwosequencesA andB. E.g.:– LCSS(AGATCTATCT,CGCCTCTATG)=TCTAT.

• Agoodsolutionistobuildsuffixtreefortheshortersequenceandmakeadescendingsuffixwalk withtheothersequence.


Suffixlink

X

aX

suffix link


Descendingsuffixwalk

suffix tree of A Read B left-to-right,always going down in thetree when possible.If the next symbol of B doesnot match any edge labelon current position, takesuffix link, and try again.(Suffix link in the root to itself emits a symbol).The node v encountered with largest string depthis the solution.v


• APL5:RecognizingDNAcontaminationRelatedtoDNAsequencing,searchforlongeststrings(longerthanthreshold)thatarepresentintheDBofsequencesofothergenomes.

• APL6: CommonsubstringsofmorethantwostringsGeneralizationofAPL4,canbedoneinlinear(intotallengthofallstrings)time


Anothercommontool:Generalizedsuffixtree

ACCTTA....ACCT#CACATT..CAT#TGTCGT...GTA#TCACCACC...C$

AC

C

node info:subtree size 47813871sequence count 87


Generalizedsuffixtreeapplication

...ACC..#...ACC...#...ACC...ACC..ACC..#..ACC..ACC...#...ACC...#...

...#....#...#...#...ACC...#...#...#...#...#...#..#..ACC..ACC...#......#...

AC

C

node info:subtree size 4398blue sequences 12/15red sequences 2/62


Casestudycontinued

genome

regions with ChIP-seq matches

suffix tree of genome

5 blue1 red

TAC..........T

motif?


• APL7:Buildingadirectedgraphforexactmatching: Suffixgraph - directedacyclicwordgraph(DAWG),asmallestfinitestateautomaton recognizingallsuffixesofastringS.Thisautomatoncanrecognize membership,butnottellwhichsuffixwasmatched.

• Construction:mergeisomorficsubtrees.• IsomorficinSuffixTreewhenexistssuffixlinkpath,andsubtreeshaveequalnr.ofleaves.


• APL8:Areverseroleforsuffixtrees,andmajorspacereductionIndexthepattern,nottree...

• Matchingstatistics.• APL10:All-pairssuffix-prefixmatchingForallpairsSi, Sj, findthelongestmatchingsuffix-prefixpair.Usedinshortestcommonsuperstringgeneration(e.g.DNAsequenceassembly),ESTalignmentmetc.


• APL11:Findingallmaximalrepetitivestructuresinlineartime

• APL12:Circularstringlinearizatione.g.circularchemicalmoleculesinthedatabase,onewantstolienarizetheminacanonicalway...

• APL13:Suffixarrays- morespacereductionwilltouchthatseparately


• APL14:Suffixtreesingenome-scaleprojects• APL15:ABoyer-Mooreapproachtoexactsetmatching

• APL16:Ziv-Lempeldatacompression• APL17:MinimumlengthencodingofDNA

ApplicationsofSuffixTrees• AdditionalapplicationsMostlyexercises...• Extrafeature:CONSTANTtimelowestcommonancestorretrieval(LCA)

Andmestruktuurmisvõimaldableidakonstantseajagaalumistühistvanemat(seevastabpikimaleühiseleprefixile!)onvõimalikkoostadalineaarseajaga.

• APL:Longestcommonextension:abridgetoinexactmatching• APL:Findingallmaximalpalindromesinlineartime

Palindromereadsfromcentralpositionthesametoleftandright.E.g.:kirik,saippuakivikauppias.

• BuildthesuffixtreeofSandinvertedS(aabcbad=>aabcbad#dabcbaa)andusingtheLCAonecanaskforanypositionpair(i,2i-1),thelongestcommonprefixinconstanttime.

• ThewholeproblemcanbesolvedinO(n).


• APL:Exactmatchingwithwildcards• APL:Thek-mismatchproblem• Approximatepalindromesandrepeats• Fastermethodsfortandemrepeats• Alinear-timesolutiontothemultiplecommonsubstringproblem

• Andmany-manymore...


Propertiesofsuffixtree

• Suffixtreehasn leavesandatmostn-1internalnodes,wheren isthetotallengthofallsequencesindexed.

• Eachnoderequiresconstantnumberofintegers(pointerstofirstchild,sibling,parent,textrangeofincomingedge,statisticscounters,etc.).

• Canbeconstructedinlineartime.


Propertiesofsuffixtree...inpractice

• Hugeoverheadduetopointerstructure:– Standardimplementationofsuffixtreeforhumangenomerequiresover200GB memory!

– Acarefulimplementation(usinglogn -bitfieldsforeachvalueandarraylayoutforthetree)stillrequiresover40GB.

– Humangenomeitselftakeslessthan1GB using2-bitsperbp.

73

1. Suffix tree

2. Suffix array


4. Finding motifs

74

Suffixes- sorted

• Sortallsuffixes.Allowstoperformbinarysearch!

hattivattiattivattittivattitivattiivattivattiattittitiiε

εattiattivattihattivattiiivattititivattittittivattivatti

75

Suffixarray:example

• suffixarray=lexicographicorderofthesuffixes



1172110594836

1234567891011

1172110594836

76

Suffixarrayconstruction:sort!

• suffixarray=lexicographicorderofthesuffixes


1172110594836

1234567891011

1172110594836

77

Suffixarray

• suffixarray SA(T)=anarraygivingthelexicographicorderofthesuffixesofT

• spacerequirement:5|T|• practitionerslikesuffixarrays(simplicity,spaceefficiency)

• theoreticianslikesuffixtrees(explicitstructure)


Reducingspace:suffixarray

AC

T

4 2 1 5 36

C A T A C T1 2 3 4 5 6

=[3,3]=[3,3]=[2,2]

suffix array

=[4,6]=[6,6]=[2,6]

=[3,6]=[5,6]

CC

TTA

T

TAC

T CT

A

T

A


Suffixarray

• Manyalgorithmsonsuffixtreecanbesimulatedusingsuffixarray...– ...andcoupleofadditionalarrays...– ...formingso-calledenhancedsuffixarray...– ...leadingtothesimilarspacerequirementascarefulimplementationofsuffixtree

• Notasatisfactorysolutiontothespaceissue.

80

Patternsearchfromsuffixarrayhattivattiattivattittivattitivattiivattivattiattittitiiε


1172110594836

att binary search


Whatwelearntoday?

• Welearnthatitispossibletoreplacesuffixtreeswithcompressedsuffixtrees thattake8.8GB forthehumangenome.

• Welearnthatbacktracking canbedoneusingcompressedsuffixarrays requiringonly2.1GBforthehumangenome.

• Welearnthatdiscovering interestingmotifseedsfromthehumangenometakes40hoursandrequires9.3GB space.

82

Recentsuffixarrayconstructions

• Manber&Myers(1990):O(|T|log|T|)• lineartimeviasuffixtree• January/June2003:directlineartimeconstructionofsuffixarray- Kim,Sim,Park,Park(CPM03)- Kärkkäinen&Sanders(ICALP03)- Ko&Aluru(CPM03)

83

Kärkkäinen-Sandersalgorithm

1.Construct the suffix array of the suffixes starting at positions i mod 3 ≠ 0. This is done by reduction to the suffix array construction of a string of two thirds the length, which is solved recursively.

2.Construct the suffix array of the remaining suffixes using the result of the first step.

3.Merge the two suffix arrays into one.

84

Notation

• stringT=T[0,n)=t0t1 …tn-1• suffixSi =T[i,0)=titi+1 …tn-1• forC\subset[0,n]:SC ={Si|iinC}

• suffixarray SA[0,n]ofTisapermutationof[0,n]satisfyingSSA[0] <SSA[1] <…<SSA[n]

85

Runningexample

• T[0,n)=yabbadabbado00…

• SA=(12,1,6,4,9,3,8,2,7,5,10,11,0)

0 1 2 3 4 5 6 7 8 9 10 11

86

Step0:Constructasample

• fork=0,1,2Bk={iє [0,n]|imod3=k}

• C=B1UB2samplepositions• SC samplesuffixes

• Example:B1={1,4,7,10},B2={2,5,8,11},C={1,4,7,10,2,5,8,11}

87

Step1:Sortsamplesuffixes• fork=1,2,construct

Rk=[tktk+1tk+2][tk+3tk+4tk+5]…[tmaxBktmaxBk+1tmaxBk+2]

R=R1^R2concatenationofR1andR2

SuffixesofRcorrespondtoSC:suffix[titi+1ti+2]…correspondstoSi;correspondenceisorderpreserving.

SortthesuffixesofR:radixsortthecharactersandrenamewithrankstoobtainR´.Ifallcharactersdifferent,theirorderdirectlygivestheorderofsuffixes.Otherwise,sortthesuffixesofR´ usingKärkkäinen-Sanders.Note:|R´|=2n/3.

88

Step1(cont.)

• oncethesamplesuffixesaresorted,assignaranktoeach:rank(Si)=therankofSiinSC;rank(Sn+1)=rank(Sn+2)=0

• Example:R=[abb][ada][bba][do0][bba][dab][bad][o00]R´ =(1,2,4,6,4,5,3,7)SAR´ =(8,0,1,6,4,2,5,3,7)rank(Si)- 14- 26- 53– 78– 00

89

Step2:Sortnonsamplesuffixes

• foreachnon-sampleSi є SB0 (notethatrank(Si+1)isalwaysdefinedforiє B0):

Si ≤Sj ↔(ti,rank(Si+1))≤(tj,rank(Sj+1))• radixsortthepairs(ti,rank(Si+1)).

• Example:S12 <S6 <S9 <S3 <S0because(0,0)<(a,5)<(a,7)<(b,2)<(y,1)

90

Step3:Merge• mergethetwosortedsetsofsuffixesusingastandardcomparison-basedmerging:

• tocompareSi є SC withSj є SB0,distinguishtwocases:

• iє B1:Si ≤Sj ↔(ti,rank(Si+1))≤(tj,rank(Sj+1))• iє B2:Si ≤Sj ↔(ti,ti+1,rank(Si+2))≤(tj,tj+1,rank(Sj+2))

• notethattheranksaredefinedinallcases!• S1 <S6 as(a,4)<(a,5)andS3 <S8 as(b,a,6)<(b,a,7)

91

RunningtimeO(n)

• excludingtherecursivecall,everythingcanbedoneinlineartime

• therecursionisonastringoflength2n/3• thusthetimeisgivenbyrecurrence

T(n)=T(2n/3)+O(n)• henceT(n)=O(n)

92

Implementation

• about50linesofC++• codeavailablee.g.viaJuhaKärkkäinen’shomepage

93

LCPtable

• LongestCommonPrefixofsuccessiveelementsofsuffixarray:

• LCP[i]=lengthofthelongestcommonprefixofsuffixesSSA[i] andSSA[i+1]

• buildinversearraySA-1 fromSAinlineartime• thenLCPtablefromSA-1 inlineartime(Kasaietal,CPM2001)

• OxfordEnglishDisctionary http://www.oed.com/• Example- WordoftheDay,Fourth

http://biit.cs.ut.ee/~vilo/edu/2005-06/Text_Algorithms/L7_SuffixTrees/wotd_fourth.htmlhttp://www.oed.com/cgi/display/wotd

• PATindex- byGastonGonnet(taonsamutiMapletarkvaraüksloojatestninghiljemmolekulaarbioloogiatarkvarapaketiväljatöötajaid)

• PATindexisessentiallyasuffixarray.Tosavespace,indexedonlyfromfirstcharacterofeveryword

• XML-tagging(orSGML,atthattime!)alsoindexed• TomarkcertainfieldsofXML,thebitvectorswereused.• Mainconcern- improvethespeedofsearchontheCD- minimizerandom

accesses.• Forslowmediumeven15-20accessesistooslow...• G.H.Gonnet,R.A.Baeza-Yates,andT.Snider,Lexicographicalindicesfor

text:Invertedfilesvs.PATtrees,TechnicalReportOED-91-01,CentrefortheNewOED,UniversityofWaterloo,1991.

95

Suffixtreevssuffixarray

• suffixtreeó suffixarray+LCPtable

96

1. Suffix tree

2. Suffix array


4. Finding motifs

97

SubstringmotifsofstringT

• stringT =t1 …tninalphabetA.• Problem:whatarethefrequentlyoccurring(ungapped)substringsofT?Longestsubstringthatoccursatleastq times?

• Thm:SuffixtreeTree(T) givescompleteoccurrencecountsofallsubstringmotifsofTinO(n) time(althoughT mayhaveO(n2)substrings!)

98

Countingthesubstringmotifs

• internalnodesofTree(T)↔repeatingsubstringsofT

• numberofleavesofthesubtreeofanodeforstringP=numberofoccurrencesofPinT

99

Substringmotifsofhattivatti


tivatti

ivatti

vatti

vattivatti

attiti

i

i

tti

ti

t

i

vatti

vatti

vatti

hattivatti

atti

2

2 2

24

Counts for the O(n) maximal motifs shown

100

FindingrepeatsinDNA

• humanchromosome3• thefirst48999930bases• 31mincputime(8processors,4GB)

• Humangenome:3x109 bases• Tree(HumanGenome)feasible

101

Longestrepeat?

Occurrences at: 28395980, 28401554r Length: 2559

ttagggtacatgtgcacaacgtgcaggtttgttacatatgtatacacgtgccatgatggtgtgctgcacccattaactcgtcatttagcgttaggtatatctccgaatgctatccctcccccctccccccaccccacaacagtccccggtgtgtgatgttccccttcctgtgtccatgtgttctcattgttcaattcccacctatgagtgagaacatgcggtgtttggttttttgtccttgcgaaagtttgctgagaatgatggtttccagcttcatccatatccctacaaaggacatgaactcatcatttttttatggctgcatagtattccatggtgtatatgtgccacattttcttaacccagtctacccttgttggacatctgggttggttccaagtctttgctattgtgaatagtgccgcaataaacatacgtgtgcatgtgtctttatagcagcatgatttataatcctttgggtatatacccagtaatgggatggctgggtcaaatggtatttctagttctagatccctgaggaatcaccacactgacttccacaatggttgaactagtttacagtcccagcaacagttcctatttctccacatcctctccagcacctgttgtttcctgactttttaatgatcgccattctaactggtgtgagatggtatctcattgtggttttgatttgcatttctctgatggccagtgatgatgagcattttttcatgtgttttttggctgcataaatgtcttcttttgagaagtgtctgttcatatccttcgcccacttttgatggggttgtttgtttttttcttgtaaatttgttggagttcattgtagattctgggtattagccctttgtcagatgagtaggttgcaaaaattttctcccattctgtaggttgcctgttcactctgatggtggtttcttctgctgtgcagaagctctttagtttaattagatcccatttgtcaattttggcttttgttgccatagcttttggtgttttagacatgaagtccttgcccatgcctatgtcctgaatggtattgcctaggttttcttctagggtttttatggttttaggtctaacatgtaagtctttaatccatcttgaattaattataaggtgtatattataaggtgtaattataaggtgtataattatatattaattataaggtgtatattaattataaggtgtaaggaagggatccagtttcagctttctacatatggctagccagttttccctgcaccatttattaaatagggaatcctttccccattgcttgtttttgtcaggtttgtcaaagatcagatagttgtagatatgcggcattatttctgagggctctgttctgttccattggtctatatctctgttttggtaccagtaccatgctgttttggttactgtagccttgtagtatagtttgaagtcaggtagcgtgatggttccagctttgttcttttggcttaggattgacttggcaatgtgggctcttttttggttccatatgaactttaaagtagttttttccaattctgtgaagaaattcattggtagcttgatggggatggcattgaatctataaattaccctgggcagtatggccattttcacaatattgaatcttcctacccatgagcgtgtactgttcttccatttgtttgtatcctcttttatttcattgagcagtggtttgtagttctccttgaagaggtccttcacatcccttgtaagttggattcctaggtattttattctctttgaagcaattgtgaatgggagttcactcatgatttgactctctgtttgtctgttattggtgtataagaatgcttgtgatttttgcacattgattttgtatcctgagactttgctgaagttgcttatcagcttaaggagattttgggctgagacgatggggttttctagatatacaatcatgtcatctgcaaacagggacaatttgacttcctcttttcctaattgaatacccgttatttccctctcctgcctgattgccctggccagaacttccaacactatgttgaataggagtggtgagagagggcatccctgtcttgtgccagttttcaaagggaatgcttccagtttttgtccattcagtatgatattggctgtgggtttgtcatagatagctcttattattttgagatacatcccatcaatacctaatttattgagagtttttagcatgaagagttcttgaattttgtcaaaggccttttctgcatcttttgagataatcatgtggtttctgtctttggttctgtttatatgctggagtacgtttattgattttcgtatgttgaaccagccttgcatcccagggatgaagcccacttgatcatggtggataagctttttgatgtgctgctggattcggtttgccagtattttattgaggatttctgcatcgatgttcatcaaggatattggtctaaaattctctttttttgttgtgtctctgtcaggctttggtatcaggatgatgctggcctcataaaatgagttagg

102

Tenoccurrences?

ttttttttttttttgagacggagtctcgctctgtcgcccaggctggagtgcagtggcgggatctcggctcactgcaagctccgcctcccgggttcacgccattctcctgcctcagcctcccaagtagctgggactacaggcgcccgccactacgcccggctaattttttgtatttttagtagagacggggtttcaccgttttagccgggatggtctcgatctcctgacctcgtgatccgcccgcctcggcctcccaaagtgctgggattacaggcgt

Length: 277

Occurrences at: 10130003, 11421803, 18695837, 26652515, 42971130, 47398125In the reversed complement at: 17858493, 41463059, 42431718, 42580925

103

Usingsuffixtrees:plagiarism

• findlongestcommonsubstringofstringsXandY

• buildTree(X$Y)andfindthedeepestnodewhichhasaleafpointingtoXandanotherpointingtoY

104

Usingsuffixtrees:approximatematching

• editdistance:insertions,deletions,changes

• STOCKHOLMvsTUKHOLMA

105

Stringdistance/similarityfunctions

STOCKHOLM vs TUKHOLMA

STOCKHOLM__TU_ KHOLMA

=> 2 deletions, 1 insertion, 1 change

106

Dynamicprogrammingdi,j = min(if ai=bj then di-1,j-1 else ¥,

di-1,j + 1, di,j-1 + 1)

= distance between i-prefix of A and j-prefix of B(substitution excluded)

di,j

di-1,j-1

di,j-1

di-1,j

dm,n

mxn table d

A

B

ai

bj

+1

+1

107

A\B s t o c k h o l m0 1 2 3 4 5 6 7 8 9

t 1 2 1 2 3 4 5 6 7 8u 2 3 2 3 4 5 6 7 8 9k 3 4 3 4 5 4 5 6 7 8h 4 5 4 5 6 5 4 5 6 7o 5 6 5 4 5 6 5 4 5 6l 6 7 6 5 6 7 6 5 4 5m 7 8 7 6 7 8 7 6 5 4a 8 9 8 7 8 9 8 7 6 5

di,j = min(if ai=bj then di-1,j-1 else ¥, di-1,j + 1, di,j-1 + 1)

dID(A,B)optimal alignment by trace-back

108

Searchproblem

• findapproximateoccurrencesofpatternPintextT:substringsP’ofTsuchthatd(P,P’)small

• dynprogrwithsmallmodification:O(mn)• lotsof(practical)improvementtricks

P

T P’

109

Indexforapproximatesearching?

• dynamicprogramming:PxTree(T)withbacktracking

P

Tree(T)

Burrows-WheelerTransformation

• BWTfortextcompressionandindexing

Burrows-Wheeler• SeeFAQ http://www.faqs.org/faqs/compression-faq/part2/section-9.html• Themethoddescribedintheoriginalpaperisreallyacompositeofthreedifferent

algorithms:– theblocksortingmainengine(alossless,veryslightlyexpansivepreprocessor),– themove-to-frontcoder(abyte-for-bytesimple,fast,locallyadaptivenoncompressivecoder)and– asimplestatisticalcompressor(firstorderHuffmanismentionedasacandidate)eventuallydoing

thecompression.

• Ofthesethreemethodsonlythefirsttwoarediscussedhereastheyarewhatconstitutestheheartofthealgorithm.Thesetwoalgorithmscombinedformacompletelyreversible(lossless)transformationthat- withtypicalinput- skewsthefirstordersymboldistributionstomakethedatamorecompressiblewithsimplemethods.Intuitivelyspeaking,themethodtransformsslackinthehigherorderprobabilitiesoftheinputblock(thusmakingthemmoreeven,whiteningthem)toslackinthelowerorderstatistics.Thiseffectiswhatisseeninthehistogramoftheresultingsymboldata.

• Please,readthearticlebyMarkNelson:• DataCompressionwiththeBurrows-WheelerTransformMarkNelson,Dr.Dobb'sJournal

September,1996.http://marknelson.us/1996/09/01/bwt/

DRD

1645*=>5240163023

DRDOBBS

0123456

1645023

CODE:t: hat acts like this:<13><10><1t: hat buffer to the constructort: hat corrupted the heap, or woW: hat goes up must come down<13t: hat happens, it isn't likelyw: hat if you want to dynamicallt: hat indicates an error.<13><1t: hat it removes arguments fromt: hat looks like this:<13><10><t: hat looks something like thist: hat looks something like thist: hat once I detect the mangled

Example

• Decode:errktreteoe.e

• Hint:. Isthelastcharacter,alphabeticallyfirst…

Lecture 14 Full Text - ut text indexing Jaak Vilo ... A suffix tree T for a string S ... • Each...

Documents

Transcript of Lecture 14 Full Text - ut text indexing Jaak Vilo ... A suffix tree T for a string S ... • Each...