1Simulation/theoryWith modestmarker spacingin a humanstudy, LOD of3 is 9% likely tobe a falsepositive.Not even counting recombinantsAAbbaaBBAaBbFlower colorinter-mateTwo loci.Need onedominant alleleat each locus toget phenotype.AAB b AaBb aaBb AaBB aaBB AabbGenotype atmarker close to Alocus32Bottomallele13TopallelewhitepurpleNOTcomplete co-inheritance.Affected sib pair method2,2 2,32,2 2,24,4 1,31,4 1,4…(1/2)*20Differentallele(1/2)*22SamealleleExpectedunder nullObservedSib pairsDoesn’t require youto know dominant orrecessive; doesn’trequire largepedigrees.Model-free (a goodthing).www.jax.org/staff/churchill/labsite/pubs/qtl.pdfUnlike cystic fibrosisand Huntington’sdisease, most traitsare not yes-or-no.E.g. blood pressure.2DistributionsDistributionsDistributionsEnvironment and error3What if…Salt waterPlain waterWhat if…Exact same mouse, every day for 6 moWhat if…Exact same mouse, every day for 6 mo• Time of day• Change in cage-mates• Age• Reproductive cycle…What if…Many clones/identical twins4What if…Many clones/identical twins• Time of day• Change in cage-mates• Age• Reproductive cycle…“Experimental error”+random variationImagine a cross:AA x aaAa(F1)Aa x AaAAAaaAaa(F2)Single-locus, dominant(F2’s)Single-locus, dominant75% 25%(F2’s)5Single-locus, dominant75% 25%AaaAAAaa(F2’s)Single-locus, dominant75% 25%AaaAAAaa(F2’s)A given genotype does notguarantee exactly the samephenotype every time.More generally:AA x BBAB(F1)AB x ABAAABBABB(F2)1 locus, ?61 locus, ?25%50% 25%1 locus, ?AAABBABB25%50% 25%1 locus, incomplete dominanceAAABBABB25%50% 25%1 locus, incomplete dominanceAAABBABB25%50% 25%Remember? Fig. 3.27AAABBABB25%50% 25%AAABBABBAnother representation25%50% 25%AAABBABBAAABBABB“Effect of having aB”1 locus, incomplete dominance25%50% 25%“Effect of having aB”AAABBABBAAABBABB1 locus, incomplete dominance25%50% 25%“Effect of having aB”AAABBABBAAABBABBEffect of a B allele is the same regardless of genotype: additive1 locus, incomplete dominance81 locus, complete dominance75% 25%ABBAAABB1 locus, complete dominance75% 25%ABBAAABB1 locus, complete dominance75% 25%ABBAAABBDominance is a kind of epistasis: nonadditiveAll these are examples of asingle locus controllingvariation in a quantitative trait.But usually…morecomplicated.9A real example(F2’s)CC x SSCSCS x CSCCSSCSSC(F2’s)A real example(F2’s)A real exampleC3H parent(F2’s)A real exampleSWR parentC3H parent(F2’s)10Not 1-locus dominant, or 1-locusincomplete dominance, or…SWR parentC3H parent(F2’s)Doesn’t look like this…Test for linkage to markers, tofind unknown geneticdeterminants, as before.Test for linkage to markers, tofind unknown geneticdeterminants, as before.Mechanics of test aredifferent.Quantitative trait linkage testInheritance at a marker(F2’s)11Quantitative trait linkage testAlleles are named after parent!(F2’s)Quantitative trait linkage testMean BP across allF2’s that carry twoSWR alleles at marker(F2’s)Quantitative trait linkage test(F2’s)Mean BP across allF2’s that carry twoC3H alleles at markerQuantitative trait linkage test(F2’s)Not countingrecombinants.Statistical test forgoodness of fit.12Locus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentAlleles are named after parent!Locus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentLocus effect(F2’s splitout bygenotype)Locus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentEach parentstrainLocus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentHomozygotesdo not look likeparent.13Locus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentHomozygotesdo not look likeparent.What do youinfer?Locus effect vs. parentsC3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentHomozygotesdo not look likeparent.What do youinfer?A single varying locus does not explain the data>1 locus controlling trait(One mouse family)What if…C3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentMagnitude oferror bars(spread withingroup) has notchanged.Locus effect isweaker.14Correct interpretation:C3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentDifferencebetween Sand C at thislocus has acausal role inbloodpressurevariation, buteffect ismodest.Correct interpretation:C3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentDifferencebetween Sand C at thislocus has acausal role inbloodpressurevariation, buteffect ismodest.“Effect of having an S allele”Correct interpretation:C3HparentF2’s, C/C atmarkerF2’s, C/S atmarkerF2’s, S/S atmarkerSWRparentMost lociunderlyinghumandisease looklike this.“Effect of having an S allele”Complex traits(one family, mouse model)15Complex traits(one family, mouse model)Complex traitsGenetic differences at both loci affect the trait(one family, mouse model)Complex traitsEach locus responsible for half?(one family, mouse model)Complex traitsEach locus responsible for half? Depends on the model.(one family, mouse model)16Complex traits(one family, mouse model)Complex traits(one family, mouse model)Each locus responsible fora third?Complex traitsIf 5 loci, each responsible for a fifth? 10 loci? …The more loci, the smaller the effects and theharder to detect.Complex traitsGenetic complexity is the rule; simple 1-or 2-locus models are the exceptionOne common result of a linkage study is nosignificant linkage anywhere.17We haven’t talked abouthumans lately…With model organisms, can always study asingle cross/family with lots of progeny, sobetter statistical power to detect weak loci.And less chance of locus heterogeneity.New York Times, Nov. 11,2007Companies offering the serviceCompanies offering the serviceCan knowing your inheritanceat SNPs predict your
View Full Document