DNA Insertions and Deletions in the Human Genome. Philipp W. Messer

Size: px

Start display at page:

Download "DNA Insertions and Deletions in the Human Genome. Philipp W. Messer"

Valentine Blankenship
10 years ago
Views:

1 DNA Insertions and Deletions in the Human Genome Philipp W. Messer

2 Genetic Variation CGACAATAGCGCTCTTACTACGTGTATCG : : CGACAATGGCGCT---ACTACGTGCATCG 1. Nucleotide mutations 2. Genomic rearrangements 3. DNA insertions / deletions (indels)

3 Outline 1. Origin and characteristics of indels 2. Indels in protein-coding regions 3. Duplications and genomic correlations 4. Duplications and alignment scores

4 Part 1 Origin and Characteristics of Indels in the Human Genome [Messer and Arndt, Mol Biol Evol 2007]

5 Identifying Insertions/Deletions Time ~5 Myr Human Chimp Rhesus High-quality flanks H:..ATACCTCGTACAATAGCGCTGGTACAGATA.. C:..ATACCTCGTACAAT--CGCTGGTACAGGTA.. R:..ATACCTCGTACAAT--CGCTGCTACAGATA.. Insertions can be distinguished from deletions (parsimony)

6 Insertion/Deletion Statistics Insertions: ~ events 15% SSR Deletions: ~ events 5% SSR

7 Molecular Mechanisms 1. Unequal Crossing Over (UCO) nonhomologous recombination 2. Replication Slippage (RS) slipped-strand mispairing

8 Indel Signatures for UCO and RS Insertions: Tandem duplications of preexisting duplicates Deletions: Remove one copy of preexisting duplicates

9 trace extension d insertion length l

10 Indel Trace Extensions UCO, RS (insertion) UCO, RS (deletion) Tandem duplication Random indel

11 Measured Trace Extensions (l=8 bp)

12 Chromosomal Rate Differences 800 Rates (bp / Mbp) Autosomes Chr X Chr Y 0 Insertions Deletions 1. Indels occur preferentially in the male germline 2. Indels are not recombination-mediated

13 Indel Characteristics 1. The majority of insertions are tandem duplications 2. Long preexisting duplicates are often missing 3. Indels occur preferentially in the male germline 4. Indels are not recombination-mediated

14 Nonhomologous End Joining DNA break End joining Small or no homology required Filling in of single strands

15 Part 2 Indels in Protein-coding Regions of the Human Genome [Chaux, Messer, Arndt, BMC Evol Biol (submitted)]

16 Indel Rates in Coding Regions

17 Genetic Code

18 Inserted/Deleted Amino Acids

19 Physico-chemical Properties

20 Physico-chemical Properties suppressed intensified

21 Protein Secondary Structure suppressed

22 Part 3 Tandem Duplications and Genomic Correlations [Messer, Arndt, Lässig, Phys Rev Lett 2005] [Messer, Lässig, Arndt, J Stat Mech 2005] [Messer, Arndt, Nucleic Acid Res 2006]

23 Sequence Evolution Model Mutation Segmental Duplication length l A C A A G T C C A rate µ rate δ l A T A A G T C C AGA T C C A Random Insertion Segmental Deletion length l A T T G T C C A rate γ l + rate γ l - A GA G A C T T A length l

24 The Correlation Function Measures likelihood of finding two G/C base pairs separated by a distance r along the genome ( ) = (, + ) 2 ( ) C r P x x r P x G/C G/C G T A T G A T C G A G A A Position: x x+r

25 Types of Correlation Behavior Random sequence Local correlations Long-range correlations Cr () 0 ( r r) Cr () exp / 0 Cr () r α Noise fluctuations Characteristic scale Generated e.g by Markov-processes Scale free, fractal Generated by a nontrivial dynamical model

26 Calculation of C(r) for Our Model Approach: continuous time Master Equation formalism Exact equation for the dynamics of C(r) Stationary solution in a continuum limit: C( r) r α with α = 4µ eff λ

27 Correlations in Genomic DNA Cr () r α distance r distance r correlation C(r) correlation C(r)

29 Part 4 Tandem Duplications and Alignment Score Statistics [Messer, Bundschuh, Vingron, Arndt, RECOMB 2006] [Messer, Bundschuh, Vingron, Arndt, JCB 2007]

30 Alignment Score Statistics Sequence alignment Seq1: ACCTAGTGCTA Significance Seq2: ATCTAGTGATA P-values of scores Requires DNA null model Standard iid model Problem Score distribution in the null model iid model correlated model Correlations in DNA Incorporate into null model P-values change e λs

31 Gaussian Approximation Analytic approach to calculate alignment score statistics for null models with LRC sequences, i.e. C(r) r -α 2 s λ = σ 2 + c ζ (2 α ) vanishes for iid sequences LRC s increases probability of finding high alignment scores by chance

32 Numerical Verification Score distribution Decay parameter λ lrc (α=0.5) lrc (α=1.0) iid Gaussian approximation captures qualitative behavior

33 Biological Significance Alignment of random sequences with same correlation parameters as human chr. 22 λ iid 1.37 λ lrc 1.21 P lrc 3x10-5 P iid 2x10-6 Difference in λ is approx.16 %, p-value increase > 10 x

34 Summary 1. The majority of short DNA insertions are tandem duplications 2. Amino acid insertions/deletions are less deleterious than substitutions 3. Tandem duplications cause long-range correlations in genomic base composition 4. These correlations have profound impact on the statistics of sequence alignment scores

35 Acknowledgments Peter Arndt Nicole de la Chaux Paz Polak Federico Squartini Ralf Bundschuh Michael Lässig Martin Vingron

Innovations in Molecular Epidemiology

Innovations in Molecular Epidemiology Molecular Epidemiology Measure current rates of active transmission Determine whether recurrent tuberculosis is attributable to exogenous reinfection Determine whether