Alhaddad Exit Seminar

Linkage Disequilibrium and Recombination in the Domestic Cat:

Applications to Genome-wide Association and Linkage Studies

Hasan AlhaddadAdvisor: Dr. Leslie Lyons

Presentation Overview

• Introduction

• Linkage disequilibrium in cats

• Population recombination rate and recombination hotspots

• Genome-wide analyses of progressive retinal atrophy in Persian cats

• Conclusion

• Acknowledgments

Introduction Linkage Disequilibrium Recombination hotspots Persian PRA Conclusion

A bit of history Cat domestication Cat breeds Phenotype genetics Disease genetics Genetic resources

“An Angora cat and a regular male cat (first generation) have produced only regular cats (second generation). If you looked at them, you would give the father a high significance. The young cats however had a lot of Angora blood despite their regular look. That is because after mating two of the same, there was in the third generation besides regular also a female Angora cat (unchanged). The chance for Angora blood may be even better in the fourth generation. We would have not reached that conclusion if we only relied on the visible traits and hence we should not trust them, because how would two regular cats produce an Angora cat?”

C. Nägeli, 1884



Personal synthesisa. Ecological domestication

b. Selection from standing variation in RB

c. Selection from standing variation in breed

d. Breeds from de novo mutation

e. Hybridizing two (more) breeds

f. Interspecies hybridization

• What is a cat breed?• Breeders vs. Genetics?

• Cat Fancy Association (CFA) recognizes 40 breeds.

• The International Cat Association (TICA) recognizes 55 breeds.

• Cat breed designation is interesting and challenging.





Filler et al., J Hered, (2012)Gandolfi et al., Scientific Reports, (2013)

Variation within FGF5 (long hair) differentiate between Persian and Exotic shorthair cats



CFA registry record of cat breeds for the period 1958-2011

(1,901,585 cats)





Oralfacial pain

Hypokalemia

Burmese

Head defect

Persian family

Polycystic Kidney

Retinal degeneration

Osteochondroplasia

Diabetes

Hypertrophic cardiomyopathy



• Somatic cell hybrid panelsO’Brien and Nash, Science 1982

• Intra-inter species linkage mapMenotti-Raymond et al., Genomics 1999 Menotti-Raymond et al., Journal of Heredity 2003Menotti-Raymond et al., Genomics 2009

• RH panelMurphy et al., Genome Research 2000Bach et al., Cytogenet Genome Res 2012

• Genome sequencingPontius et al., Genome Research 2007 – 2X ~ 60% genomeMullikin et al., BMC Genomics 2010 – 3X ~ 80% genome and 1,000,000 SNPs

• 63K Illumina Array chip



O1: Estimate the extent of LD in cats.

O2: Provide insights for design of GWA studies.

H1: Linkage disequilibrium varies across populations and genomic regions.

H2: Variation in linkage disequilibrium among populations reflects population histories.

Objectives Hypotheses


Linkage Disequilibrium Markers Samples LD in cats LD in mammalsHaplotypes Cat LD & GWAS

• Linkage disequilibrium is the non-random association of alleles at different loci in a gamete.


Markers Samples LD in cats LD in mammalsHaplotypes Cat LD & GWAS

• LD is measured as the difference between the frequency of a haplotype and the frequencies of the alleles.

D` =DAB , where Dmax = {

min (PA PB, Pa Pb), when DAB < 0

Dmax min (PA Pb, Pa PB), when DAB > 0

r2 = D2

PA Pa PB Pb

• Squared correlation coefficient

• Normalized D` (D prime)

DAB = PAB – PAPB

Linkage Disequilibrium



Hill and Weir, Theor Popul Biol 33, 54 (1988)

LD decay and comparison across populations




• A custom Illumina GoldenGate array (1536 SNPs).

• Ten (1 Mb) regions from various locations relative to centromere.

• ~ 150 SNPs/region with higher density of marker in one end of the region.




18 breeds

2 random bred populations

Total of 408 samples

~ 18 cats/population

Breed cats pedigreed verified to be unrelated to the grandparent level




Extent of LD in cats on each chromosomal regionLinkage Disequilibrium



• Lowest LD among breeds found in Siberian and Manx breed (~ 20Kb). These breeds largely resemble a random bred population.

• Highest LD (> 200 Kb) found in Eastern breeds (Birman, Burmese, and Siamese). These breeds are known to have restricted phenotypes and under selection for very defined phenotypes.

• Persian cats exhibit moderate LD (~ 75 Kb) compared to other breeds and is consistent with the large population size of the breed.



Markers Samples LD in cats LD in mammalsHaplotypes Cat LD & GWASLinkage Disequilibrium



• Eastern Breeds have higher fractions of SNPs useful for GWA studies.

• Association studies for phenotypic vs. disease traits.

• Successes so far.



Markers Samples LD in cats LD in mammalsHaplotypes Cat LD & GWASLinkage Disequilibrium

• LD calculated using

the same measure.

• Extent of LD

compared at the same

value.

• Each data point

corresponds to the LD

of a single population.


LD Markers Samples LD in cats LD in mammalsHaplotypes Cat LD & GWAS

Conclusion

Any questions?

• Extent of LD varies across chromosomal localities and among breeds.

• LD of cat breeds reflects the demographic and breeding histories.

• Eastern breeds exhibit larger extent of LD. Association studies in involving such breeds are likely to be successful.

• Breeds such as Manx and Siberian (may be Persian) require a higher density SNP array or alternative approaches.


A glance at recombination hotspots in the domestic cat

Alhaddad, H., Zhang C., Rannala B., and Lyons L.A.


O: Use coalescent based methods to infer population parameters.

O: Understand recombination hotspots in cats.

H1: Population recombination rates vary at different genomic localities.

H2: Variation in population recombination rate at different regions is due to the presence of specific genomic features (i.e. recombination hotspots)



CoalescentRecombination Markers samples program Hot&warm-spots G. featuresEstimates&interpretation

Parent

Recombination

• Recombination: shuffling the genome during meiosis

and formation of gametes.

• Recombination hotspots: regions of elevated

recombination rate.

• Correlation with hotspots in humans:

• GC content

• Repeat elements

• Motif

• PRDM9-zinc finger domain

• Dog hotspots different !



Rosenberg & Nordborg, Nat Rev Genet 3, 380 (2002)

• Coalescent + recombination = Ancestral Recombination Graph (ARG).

• Allows estimation of population parameter (rho) = scaled recombination rate.

• Identify recombination hotspots.

• Overcome the need for large pedigrees and sperm typing.



• 22 Eastern random bred (feral) cats.

• 701 SNPs distributed over ten regions.

• inferRho: coalescent-based Bayesian method.

• Bayesian hypothesis testing and Bayes factor:

• Spot designation:

• Bayes factor ≥ 100 “hot spot”

• Bayes factor 10 – 100 “warm-spot”

• Bayes factor < 10 “neutral spot”

Wang and Rannala, Philos Trans R Soc Lond B Biol Sci 363, 3921 (2008)Wang and Rannala, Proc Natl Acad Sci U S A 106, 6215 (2009)

)1()1(

ii

ii

qqppBF



Analysis of chromosome E2 region• Hotspot:

• Local recombination rate.

• Posterior probability.

• Bayes Factor.

• Four recombination hotspots in three regions.

• hotspot size: 1.8 – 4.6 Kb.

• 52 warm-spots found in all regions.

• Warm-spot size: 0.4 – 5.6 Kb.



Lack of significant correlation between GC content and hot/warm-spot locality



Hotspots Warm-spots Neutral-spots

• L2 LINE elements were present in three of the four hotspot regions.

• tRNA-Lys family SINE elements are present in three of the four hotspots.

• MIR family SINE elements were present in all hotspot regions.

• Other repeat elements inconsistently present across the hotspot regions.


Conclusion

Any questions?

• Population recombination rate varies across regions examined.

• Regions with higher recombination rates contain hotspots.

• Four recombination hotspots were identified.

• No correlation between GC content and hotspot locality (need more data).

• L2, MIR, tRNA-Lys likely to be a signature of hotspots in cats (need more data).


Localization of progressive retinal atrophy of Persian cats using genome-wide

analyses

Alhaddad, H., Gandolfi B., Grahn R.A., Rah H., Peterson C.B., Maggs D.J., Pedersen N.C., and Lyons L.A.



O: Evaluate genome-wide methods for cat pedigree data.

O: Identify the causative mutation of PRA in Persian cats.

H: Progressive Retinal Atrophy (PRA) of Persian cats can be localized using a dense genotype data of a family.

H: The causative mutation can be identified using candidate gene approach.


PRA TDT Candidate genesSamples Linkage Case-Control Haplotypesib-TDT

Rah et al., Invest Ophthalmol Vis Sci 46, 1742 (2005)

• Two identified mutations cause retinal degeneration in Abyssinian cats.

• CEP290

• CRX

• Progressive Retinal Atrophy in Persian cat.• Gradual degeneration of photoreceptor.

• Early onset: starts 2-3 weeks of age.

• Rapid disease progression.

• Complete loss of photoreceptors at 16-17 weeks of age.

• Autosomal recessive mode of inheritance.



• Pedigree composed of 202 cats.

• 126 cats (blue and red) were genotyped using 63K SNP array.

• After QC 106 included (37 affected and 69 unaffected).



• Two variation of nonparametric linkage analysis – not significant

• Recessive mode of inheritance – parametric analysis.

• Parametric analysis: 35 markers LOD score ~ 14.

• Linkage found on cat chromosome E1 (~1.75 Mb).



TDT

sib-TDT

Spielman et al., Am J Hum Genet 52, 506 (1993)Spielman, and Ewens, Am J Hum Genet 62, 450 (1998)

• Family based methods tests linkage given the presence of association between a marker locus and a disease locus.

• TDT: 33 trios.

• sibTDT: 85 sib pairs



• Case-control using pedigree data is inappropriate.

• Population substructure within the pedigree.

• Reduction of genomic inflation (3.18 to 1.3).

• Association consistent with TDT and sibTDT.



• Association and linkage markers points to a single haplotype among affected cats.

• Three control cats share the same haplotype of the affected.

• After examination two founder cats confirmed affected.

• One cat mislabeled as unaffected.



• Haplotype region contains 22 eye related genes.

• Mutation within PITPNM3, AIPL1, and ARRB2 known to cause retinal degeneration in humans.



Conclusion

Any questions?

• Parametric linkage and various association analyses consistently points to the same region.

• Progressive Retinal Atrophy of Persian cats was localized to ~ 1.36Mb region on cat chromosome E1.

• Three genes are likely candidates.


Future Direction

• Genome-wide estimation of LD in cats using 63K array.

• Investigating recombination across the genome and studying genomic signatures of recombination hotspots.

• Comparing recombination hotspots across different cat populations.

• Investigating PRDM9 in cats and analyzing the motif recognition portion.

• Studying PRA localized region via candidate gene sequencing, targeted sequencing, or whole genome sequencing.

• Continue on the development of cat genetic resources.

AcknowledgmentsDissertation Committee• Dr. Leslie A. Lyons (Chair)• Dr. Bruce Rannala• Dr. Jeffrey Ross-IbarraLyons Laboratory• Dr. Barbara Gandolfi• Dr Robert A. Grahn• Razib Khan• Many grad students• Many Many cool undergradsCCAH• Dr. Niels Pederson and lab• Dr. Ben Sacks and lab• Dr. Holly Ernest and lab

Collaborator• Dr. Chi Zhang GGG• Carolyn Yrigollen• Gavin Rice•Lattha Souvannaseng• Rebecca Nitcher• My cohort (Fall 2009)• All GGG students• Demian Sainz• Ellen PichtVet genetics lab• All are cool