Mapping by recurrence and modelling the mutation rate

Transcription

1 Current knowledge is from apping by recurrence and modelling the mutation rate Shamil Sunyaev Broad Institute of.i.t. and Harvard Comparative genomics Experimental systems: yeast reporter assays Potential problems De novo and somatic mutations 1) Selection and biased gene conversion influence sequence divergence 2) Demography of ancestral populations 3) utation rate may evolve fast (recombination rate does) De novo germ line mutations other Father Child Somatic mutations Somatic cell Somatic cell De novo mutations in statistical genetics Is it possible to find genes involved in complex traits using de novo mutations? Possible approaches that do not involve controls Estimate genomic mutation rate using independent samples Is it possible to find all drivers of cancer progression using the analysis of somatic mutations? Evaluate probability to observe recurrent events in a given gene Correct for multiple testing ajor principle: find genes with RECURRENT events PROBLE: heterogeneity among samples

2 Possible approaches that do not involve controls Real data utation rate is variable along the genome Random permutations Replication fidelity DN damage DN repair CpG deamination Regional variation of mutation rate PROBLE: heterogeneity of mutation rate along the genome Context dependence of mutation rate The rate and the effect of mutations are not statistically independent Can mutation rate heterogeneity evolve? important sites TCGTTCGGTCGCCGCGGCGTCG TTGTTCGTCGCCGCGGCTTG CpGs become mutable Important sites are more mutable sexual population efficient selection on mutation rate gene influencing mutation rate (mutation rate modifier) all active genes This is also true for individual amino acid sites CCG->CTG CCC->CTC P->L P->L utation rate 14.5 Polymorphism 11.2 Divergence 7.1 Schmidt, Gerasimova et al., PLoS Genetics 28 Can mutation rate heterogeneity evolve? Can mutation rate heterogeneity evolve? sexual population efficient selection on mutation rate Sexual population inefficient selection on mutation rate

3 Can mutation rate heterogeneity evolve? Cis-acting modifiers of mutation rate Sexual population inefficient selection on mutation rate Cis-acting modifiers of mutation rate Germ line mutation rates are associated with replication timing The model predicts a specific scale of mutation rate variation The model predicts that genes and regulatory elements would be enriched in regions of low local mutation rate Stamatoyannopoulos, dzhubei et al., Nature Genetics 29 Genome of Netherlands Consortium (GoNL) Detecting de novo mutations ZH 168 NH 91 UT 48 FL FR 62 GD 57 GR 57 DR 56 OV 58 ZL 46 NB 68 LI 58 GoNL

4 Validation results on 843 sites Context dependency mutation frequencies (xyz>xwz) 1. Random Forests.75 1 Sensitivity.5.25 PhaseByTransmission predicted relative mutation density 1 factor(cpg) CpG non CpG mutation C>T >G C>G G>T >C >T Specificity yriads of mutations were simulated at the read level uniformly along the genome were used to account for false-negatives expected relative mutation rates Functional regions are more mutable Complementary mutations 1) utation rate is significantly elevated in exons. 2) utation rate is significantly elevated in regulatory regions marked by DNase I hypersensitivity. 3) Both effects are fully explained by sequence context. >G and T>C (:T > G:C) C>T and G> (C:G > T:) G>T and C> (G:C > T:) >T and T> (:T > T:) C>G and G>C (C:G > G:C) >C and T>G (:T > T:) Transcription is asymmetric Symmetry breaking by transcription letter Transcription-associated mutational asymmetry in mammalian evolution Phil Green 1, Brent Ewing 1, Webb iller 2, Pamela J. Thomas 3, NISC Comparative Sequencing Program 3,4 & Eric D. Green 3,4 Published online 3 arch 23, doi:1.138/ng113 Proportion of mutations p = 3.5x1-4 Strand Transcribed Non-transcribed >G G> >C >T C> C>G Transitions Transversions

5 Observed inter DN distance (log1 bp) Clustering of mutations along the genome 8 utations Within individuals 6 Between individuals Expected inter DN distance (log1 bp) 8 Clustering of mutations along the genome Enrichment of C>G changes in clusters less than 2kb? DNs Proportion of DNs.5 p = 2.5x1-5 DN clusters.4 p = 3.6x1-13 p = C:G>T: Transitions C:G>:T :T>C:G :T>T: C:G>G:C Transversions Variation in nucleotide diversity Weak association between mutation rate and recombination rate Begun-quadro effect Recombination map :T>G:C Genethon arshfield decode (22) decode (21) Hapap CEU Effect estimate.3.5

6 utation vs. recombination Regional variation (1b scale) 1) Recombination is not a strong determinant of mutation. 2) utation rate and recombination rate independently influence nucleotide diversity. 3) This means that forces other than mutation underlie Begun-quadro effect. Letter Genome nalysis Patterns in spontaneous mutation revealed by human baboon sequence comparison Joana C. Silva and lexey S. Kondrashov The scale of mutational variation in the murid genome Daniel J. Gaffney 1 and Peter D. Keightley Institute of Evolutionary Biology, shworth Laboratories, School of Biological Sciences, Un Regional variation in light of species divergence Comparative genomics vs. de novo mutations Density Rates based on simulations Uniform model Comparative genomics model Observed mutation rate djusted mutation rate for recombination 1) Comparative genomics models capture only about half of variation in mutation rate 2) Forces other than mutation substantially influence human-chimpanzee divergence Correlation with rates from comparative genomics model.5 Dependence on replication timing and the effect of paternal age Fraction of genic mutations depends on paternal age

7 utation vs. replication timing Conclusion utation rate is variable along the genome 1) The preference of mutations for late replicating regions is weaker in older fathers 2) The proportion of mutations in genes is greater in children of older fathers Functional regions of the genome may exhibit mutation rate different from non-functional regions Statistical tests for recurrent mutations should be based on an accurate model of local mutation rate