Repeats!

1
Repeats!
2
Introduction

• A repeat family is a collection of repeats which
  appear multiple times in a genome.

• Our objective is to identify all families of
  interspersed repeats within a single genome

3
Challenges when identifying repeat families
. . .
. . .

• Challenges
• Regions containing repeat occurrences are not
  known a priori
• Repeat boundaries are not known a priori
• Many repeat occurrences appear as partial copies

4
Why are repeats important

• Repeats have been implicated in
• Genome rearrangements (Kazazian, 2004 Achaz et
  al 2003)
• Accelerated loss of gene order (Rocha et al,
  2003)
• Creation of novel biological functions (Lynch et
  al, 2002)
• Increased rate of evolution under stress (Capy et
  al, 2000)

5
Identifying repeats de novo

• Assume we get a new genome and we know nothing
  about it, we can
• Use a database of known repeats
  (RepeatMasker/RepBase)
• novel repeat elements may not be in the database
• repetitive gene families are never in the
  database
• Identify repeats de novo using sequence analysis

6
Existing methods for detection of repeat families

• Nearly all existing algorithms for de novo
  identification of repeat families rely on a set
  of pairwise similarities

• REPuter (Kurtz et al., 2000)
• RepeatFinder (Volfovsky et al., 2001)
• RECON (Bao and Eddy, 2002)
• RepeatGluer (Pevzner et al., 2004)
• PILER (Edgar and Myers, 2005)
• RepeatScout (Price et al, 2005)

7
Mutational forces at play

• Over time, indels substitutions will affect
  copies of repeat families
• AGGCTACCCCTTTAGGCTAGGGGGGAGGCTATCTCTCCTAGGCTATTTTT
  TAGCCTATT
• AGGCTGCCCCTTTAGGCTDGGGGGGAGGCTATCTCTCCTAGGCTATTTTT
  TAGCCTATT
• AGGCTGCCCCTTTAGGCTGGGGGGAGGCTCTCTCTCCTAGCCTATTTTTT
  AGCCTATT
• AGGCTGCCCCTTTAGGCTGGGGGGAGGCTCTCTCTCCTAGCCTATTTTTT
  AGCDTATT
• AGGCTGCCCCTTTAGGCTGGGGGGAGGCTCTCTCTCCTAGCCTATTTTTT
  AGCTATT
• AGGCTGCCCCTTTAGGCTGGGGGGAGGCTCTCTCTCCTAGCCTATTTTTT
  AGCTATT
• Require alignments ( gaps) to attempt to
  reconstruct true repeat boundaries

8
de novo repeat detection

• One approach self-search with a pairwise
  local-alignment tool such as BLAST
• Number of pairwise alignments grows O(r2) in the
  copy number of the repeat
• Inherent difficulty defining repeat boundaries
  among collections of pairwise alignments

9
Alternative methods?

• Local multiple alignment

A single local multiple alignment uses O(N) space
for a genome of length N
10
Local multiple alignment

• Local multiple alignment has the inherent
  potential to avoid pitfalls associated with
  pairwise alignment.
• But multiple alignment under the SP objective
  function remains intractable
• Progressive alignment heuristics offer excellent
  speed and accuracy (i.e. MUSCLE).
• So why not directly construct a multiple
  alignment?

11
(No Transcript)
12
Steps 1-3 Chaining seeds from the Input Sequence

• The method incorporated three novel ideas
• (1) palindromic spaced seed patterns to match
  both DNA strands simultaneously
• (2) seed extension (chaining) in order of
  decreasing multiplicity, and
• (3) procrastination when low multiplicity matches
  are encountered.

13
Step 4 Gapped Extension

• After chaining a seed match, we must perform
  gapped extension to approximate the true repeat
  boundaries
• This is an essential step to consider, assuming
  we would like to improve repeat boundary
  predictions
• But how can this be done efficiently?

14
Our approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA GTTGCGGCCCCTGAGGCACTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA CTTAAGGCCCCTGAGGATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGGTAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG CCCGAGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA TAAGCGGCCCCTGAGGCACTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC ATTGGGGCCCCTGAGGA
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGCCAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG CCCACGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . GATTCGGCCCCGAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
15
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA GTTGCGGCCCCTGAGGCACTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA CTTAAGGCCCCTGAGGATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGGTAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG CCCGAGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA TAAGCGGCCCCTGAGGCACTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC ATTGGGGCCCCTGAGGA
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGCCAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG CCCACGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . GATTCGGCCCCGAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Dynamically calculate extension window 70e
-0.01Mi Mi 200 , l 10
16
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA GTTGCGGCCCCTGAGGCACTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA CTTAAGGCCCCTGAGGATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGGTAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG CCCGAGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA TAAGCGGCCCCTGAGGCACTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC ATTGGGGCCCCTGAGGA
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGCCAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG CCCACGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . GATTCGGCCCCGAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Use MUSCLE to perform alignment of extension
window
17
HMM approach to gapped extension
ACAAGGGCCC-TGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTC
CGACTAGCAGCCA TACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CT
GGTCCTTTCCTTTAATTTGACATGA TTCATCCCCCCTGAGG-ATCTCTT
TAACGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCC-TCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGA
CAGGATGGACG AGGCCGGCCC-TGAGGTATCTCTTTAACGTTTC-CTGG
TCCTTTCCTTAAAAAAATTAAAA AACCCGCCCCCTGAGGCA-CTCTTTA
ACGTTTA-CTGGTCCTTTCCAATTTGCTCTATGTC ATTTTGCCCCCTGA
GG-ATCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GG
AAAGCCCC-TGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAG
CGCCCGCGG ATTCCGCCCC-TGAGGCATCTCTTTAACGTTTC-CTGGTC
CTTTCCGACCGAATTAATTAA . . . ATTCGGCCCC-CGAGGCATCTC
TTTAACGTTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Use HMM to detect unalign unrelated sequence
18
HMM approach to gapped extension
ACAAGGGCCC-TGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTC
CGACTAGCAGCCA TACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CT
GGTCCTTTCCTTTAATTTGACATGA TTCATCCCCCCTGAGG-ATCTCTT
TAACGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCC-TCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGA
CAGGATGGACG AGGCCGGCCC-TGAGGTATCTCTTTAACGTTTC-CTGG
TCCTTTCCTTAAAAAAATTAAAA AACCCGCCCCCTGAGGCA-CTCTTTA
ACGTTTA-CTGGTCCTTTCCAATTTGCTCTATGTC ATTTTGCCCCCTGA
GG-ATCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GG
AAAGCCCC-TGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAG
CGCCCGCGG ATTCCGCCCC-TGAGGCATCTCTTTAACGTTTC-CTGGTC
CTTTCCGACCGAATTAATTAA . . . ATTCGGCCCC-CGAGGCATCTC
TTTAACGTTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Extension successful, continue extending
19
HMM approach to gapped extension
ACAAGGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA ACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA TCATCCCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG AGGCCGGCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA ACCCGCCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC TTTTGCCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG ATTCCGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . ATTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
20
HMM approach to gapped extension
ACAAGGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA ACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA TCATCCCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG AGGCCGGCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA ACCCGCCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC TTTTGCCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG ATTCCGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . ATTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Use HMM to detect unalign unrelated sequence
21
HMM approach to gapped extension
ACAAGGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA ACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA TCATCCCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG AGGCCGGCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA ACCCGCCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC TTTTGCCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG ATTCCGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . ATTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
Finished leftward extension, now to the right
22
HMM approach to gapped extension
ACAAGGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GACTAGCAGCCA ACGAGCCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCCTTTCCTTTAATTTGACATGA TCATCCCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCCTTTCCAAGAGCCCCCGTAGC
AGAACGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCCGAGAC
AGGATGGACG AGGCCGGCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
CTTTCCTTAAAAAAATTAAAA ACCCGCCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCCTTTCCAATTTGCTCTATGTC TTTTGCCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCCTTTCCGGCCCTTATAGGTAC GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCCTTTCCAAAGAGCGCCC
GCGG ATTCCGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTCC
GACCGAATTAATTAA . . . ATTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCCTTTCGTTTCCCCCCGGCCCC
23
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GA---GCAGCCA TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCC---TTTCCTTTAATTTGACA TTCATGCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCC---TTTCCAAGAGCCCCCGT
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCGAG
ACTAGGATGG CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
C---TTTCCTTAAAAAAATTA AACCCGCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCC---TTTCCAATTTGCTCTAT TTTTTGCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCC---TTTCCGGCCCTTATAGG GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCCAAAGAGCG
CCCG CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCC---TT
TCCGACCGAATTAAT . . . -TTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCC---TTTCGTTTCCCCCCGGC
Perform MUSCLE alignment on window
24
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GA---GCAGCCA TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCC---TTTCCTTTAATTTGACA TTCATGCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCC---TTTCCAAGAGCCCCCGT
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCGAG
ACTAGGATGG CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
C---TTTCCTTAAAAAAATTA AACCCGCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCC---TTTCCAATTTGCTCTAT TTTTTGCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCC---TTTCCGGCCCTTATAGG GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCCAAAGAGCG
CCCG CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCC---TT
TCCGACCGAATTAAT . . . -TTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCC---TTTCGTTTCCCCCCGGC
Use HMM to detect unalign unrelated sequence
25
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GA---GCAGCCA TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCC---TTTCCTTTAATTTGACA TTCATGCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCC---TTTCCAAGAGCCCCCGT
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCGAG
ACTAGGATGG CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
C---TTTCCTTAAAAAAATTA AACCCGCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCC---TTTCCAATTTGCTCTAT TTTTTGCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCC---TTTCCGGCCCTTATAGG GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCCAAAGAGCG
CCCG CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCC---TT
TCCGACCGAATTAAT . . . -TTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCC---TTTCGTTTCCCCCCGGC
Extension successful, continue extending
26
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
GAGCAGCCACCA TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CTGG
TCC---TTTCCTTTAATTTGACA TTCATGCCCCTGAGG-ATCTCTTTAA
CGTTTCTCTGGTCC---TTTCCAAGAGCCCCCGT
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCGAG
ACTAGGATGG CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGGTC
C---TTTCCTTAAAAAAATTA AACCCGCCCCTGAGGCA-CTCTTTAACG
TTTA-CTGGTCC---TTTCCAATTTGCTCTAT TTTTTGCCCCTGAGG-A
TCTCTTTAACGTTTA-CTGGTCC---TTTCCGGCCCTTATAGG GGAAAG
CCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCCAAAGAGCG
CCCG CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCC---TT
TCCGACCGAATTAAT . . . -TTCGGCCCC-GAGGCATCTCTTTAACG
TTTA-CTGGTCC---TTTCGTTTCCCCCCGGC
Use MUSCLE to perform alignment of extension
window
27
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
---GAGCAGCCAC- TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CT
GGTCC---TTTCCTTTAATTTGA---- TTCATGCCCCTGAGG-ATCTC
TTTAACGTTTCTCTGGTCC---TTTCC----AAGAGCCCCC
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCG--
-AGACTAGGAT- CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGG
TCC---TTTCCTTAAAAAAAT---- AACCCGCCCCTGAGGCA-CTCTTT
AACGTTTA-CTGGTCC---TTTCC---AATTTGCTCT- TTTTTGCCCCT
GAGG-ATCTCTTTAACGTTTA-CTGGTCC---TTTCC----GGCCCTTAT
A GGAAAGCCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCC
---AAAGAGCGCC- CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CT
GGTCC---TTTCC----GACCGAATTA . . . -TTCGGCCCC-GAGGC
ATCTCTTTAACGTTTA-CTGGTCC---TTTCG----TTTCCCCCCG
Use HMM to detect unalign unrelated sequence
28
HMM approach to gapped extension
TAGTTGGCCCTGAGGCATCTCTTTAACGTTTC-CTGGTCCTTTTTTCC
---GAGCAGCCAC- TACGAGCCCCTGAGGCA-CTCTTTAACGT-TA-CT
GGTCC---TTTCCTTTAATTTGA---- TTCATGCCCCTGAGG-ATCTC
TTTAACGTTTCTCTGGTCC---TTTCC----AAGAGCCCCC
AGAAAGGCCCTCAGGCATCTCTTTAACGTTTC-CTGGTCC---TTTCG--
-AGACTAGGAT- CCGATGCCCCTGAGGTATCTCTTTAACGTTTC-CTGG
TCC---TTTCCTTAAAAAAAT---- AACCCGCCCCTGAGGCA-CTCTTT
AACGTTTA-CTGGTCC---TTTCC---AATTTGCTCT- TTTTTGCCCCT
GAGG-ATCTCTTTAACGTTTA-CTGGTCC---TTTCC----GGCCCTTAT
A GGAAAGCCCCTGAGGCATCTCTTTAACGTTTCTCTGGTCC---TTTCC
---AAAGAGCGCC- CCTATGCCCCTGAGGCATCTCTTTAACGTTTC-CT
GGTCC---TTTCC----GACCGAATTA . . . -TTCGGCCCC-GAGGC
ATCTCTTTAACGTTTA-CTGGTCC---TTTCG----TTTCCCCCCG
Extension failed, stop extending
29
Wait a moment..

• The MUSCLE alignment software reports the highest
  scoring global multiple alignment of the input
  sequences, regardless of common ancestry.
• As a result, it is likely that this method
  forcibly aligns unrelated sequence.
• HMMs to detect alignments of unrelated sequence.

30
Step 5 detecting unrelated sequence

• The HMM consists of two hidden states, Homologous
  and Unrelated.
• The observable states are the pairwise alignment
  columns, which are all possible pairs in
  A,G,C,T,- with strand and species symmetry
• i.e. AGGATCCT.
• The emission probabilities for each possible pair
  of aligned nucleotides were extracted from the
  HOXD substitution matrix presented by Chiaromonte
  et al.

31
0.5
UUUU
U
H

• Compute emission frequencies for the Unrelated
  state of our HMM using the background frequencies
  of G/C and A/T, assuming strand and species
  symmetry
• UAA UAT UTA UTT (fAT)/2 (fAT)/2
• UCC UCG UGC UGG (fGC)/2 (fGC)/2
• UAC UAG UTC UAG (fAT)/2 (fGC)/2
• UCA UCT UGA UTT (fGT)/2 (fAT)/2

32
0.5
UUUUUU
H
UU

• To empirically estimate gap-open and extend
  values for the unrelated state, align a 10-kb,
  48 GC content region taken from E. coli CFT073
  (Accession AF447814.1, coordinates 37,300-38,300)
  with an unrelated sequence.

33
0.5
UUUUUUUUUUUU
H
UU

• Alignment with MUSCLE on unrelated sequence and
  counted the number of gap-open and gap-extend
  columns in the alignment of unrelated sequences.

34
0.5
UUUUUUUUUUUUUUUUUUHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
H
UU

• Gap-open and extend frequencies for the
  homologous state were estimated by constructing
  an alignment of 10kb of orthologous sequence
  shared among a pair of divergent organisms.

35
0.5
UUUUUUUUUUUUUUUUUUHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
H
UU