Lection 2: Genetic variation in transcription networks Kristy Harmon, Lauren McIntyre, Marta Wayne, UF Gainsville, UF Gainsville Matt Hahn, Linda Bisson, Kristy Harmon IU Bloomington
Evolution in Levels Molecular Evolution Quantitative Genetics Evolution of Phenotype
Intraspecific Transcriptome Variation Components of Variation Dimensionality of Variation Phenotypic Effects?
Affymetrix microarrays:13,966 genes Heritability of expression: 663 genes: cis + trans: 8.7% trans: 6.8% P= Intraspecific Variation
Male-Female Differences in Splicing v1. Genetics v2. ? v3. Genome Biology
Male-Female Differences in Splicing Total Above Line & sex Line Sex Probe type genes control significance signif. signif. Alter. Trans Gene family Singleton Total % of “above control” n/a 57% 3% 53% Probe type Above Line, sex & probe Line&probe Sex&probe control significance significance significance Altern. Trans Gene family Total % of “above control” 23% 3% 21%
Intraspecific Transcriptome Variation: Components of Variation ,21,31,41,51,61,71,81,9 22,12,32,42,52,62,72,82,9 33,13,23,43,53,63,73,83,9 44,14,24,34,54,64,74,84,9 55,15,25,35,45,65,75,85,9 66,16,26,36,46,56,76,86,9 77,17,27,37,47,57,67,87,9 88,18,28,38,48,58,68,78,9 99,19,29,39,49,59,69,79,8 Sires Dams
Diallel data summary Agilent 60b oligos 9,863 genes were included in analysis (requirements: no multiple splice products, and probes higher than negative controls) –1,609 X linked genes –8,213 “autosomal” genes (2, 3 only) –35 4th chromosome –6 Y chromosome Models run for sexes separately; evaluated significance at FDR = 0.20 P = 2 GCA + SCA + 2 RGCA + RSCA +
Conclusions: more genes genetically vary for transcript level in females MaleFemale XAutosomeX Autosome Genetic GCA SCA rGCA rSCA
GCA, SCA ,21,31,41,51,61,71,81,9 22,12,32,42,52,62,72,82,9 33,13,23,43,53,63,73,83,9 44,14,24,34,54,64,74,84,9 55,15,25,35,45,65,75,85,9 66,16,26,36,46,56,76,86,9 77,17,27,37,47,57,67,87,9 88,18,28,38,48,58,68,78,9 99,19,29,39,49,59,69,79,8 Sires Dams
Conclusions: but more genes have heritable variation in transcript level in males MaleFemale XAutosomeX Autosome Genetic GCA SCA rGCA rSCA (and heritable variation is underrepresented on X)
rGCA; rSCA ,21,31,41,51,61,71,81,9 22,12,32,42,52,62,72,82,9 33,13,23,43,53,63,73,83,9 44,14,24,34,54,64,74,84,9 55,15,25,35,45,65,75,85,9 66,16,26,36,46,56,76,86,9 77,17,27,37,47,57,67,87,9 88,18,28,38,48,58,68,78,9 99,19,29,39,49,59,69,79,8 Sires Dams
Conclusions: males possess ~no epistatic variation in transcript levels, while females are overwhelmed with it MaleFemale XAutosomeX Autosome Genetic GCA SCA rGCA rSCA (and it is overrepresented on the X chromosome)
Overall “conclusions” Heritable variation in transcript level is “consistent” with mutation-selection balance; Dominance / epistatic variation might (?) be consistent with “antagonistic arms race”. Benefit to malesfemales Harmful for femalesmales Dominance condition recessivedominant
Intraspecific Transcriptome Variation Components of Variation Dimensionality of Variation Phenotypic Effects?
Intraspecific Transcriptome Variation: Phenotypic Effects. 9 lines eggs (5+8h) Affymetrix chips.
Binding sites of segmentation genes (from Schroeder et al. 2004)
Model Model Fit Parameter Estimate Pr > F / Significance (P) oc = bcd gt kr kni oc = tor gt kni ems = kni < {btd}* cnc = Tor Bcd Gt Kr Kni {hb} Kr = Gt Kni Cad Gt = Bcd {Gt} Kr Kni {} Kni = Tor Bcd Kr {eve} h = Kr < h = Kr Kni Cad h = Gt Cad run = Bcd Kr Kni ftz = Bcd Gt Cad = Gt Kni {Cad} {} {spl2} {nub} pdm2 = Bcd Kr Cad D = Bcd Gt Kr Kni Cad {hb} {eve} Kr = Bcd Gt Kni Cad h = Gt Cad Kr = Tor Bcd Gt Kni h = Bcd Kr Kni Cad Gt = Kr Kni h = Kr Kni Cad {eve} h = Kr Kni run = Tor Bcd Kr Kni Cad odd = Bcd Kr Cad < run = Kr Cad Gt = Kr Cad run = Bcd Kr Kni Cad Kni = Bcd Kr {Kni} {} odd = Gt Kr Kni Cad <
Factor Analysis for Expression Data Factor is a linear combination Measurements “load” on of measurements factors In every genotype, the value of the factor can be calculated and correlated with the trait value genes with high “loads”
How many dimensions? (among 9 genotypes)
Flies Yeast No tissue problems; Bunch of phenotypes ethanol; temperature; sulfates; …. 30 genotypes (4 reps including dye swaps); Log phase; Agilent arrays.
Does variation in TF expression level account for variation in expression of targets? ADR1 Transcription Factor ABF1 ADR1 AFT2 CHA4 CRZ1 CTH1 DAL80 DAL80 FAP1 FHL1 FKH1 FKH2 FZF1 GAT2 GAT3 GAT4 GIS1 GIS2 GLN3 GLN3 GZF3 HAC1 HMS1 HSF1 IFH1 MAL13 MAL33 MBP1 YER028C MSN2 MSN4 NDT80 PHO2 PHO2 PHO4 PLM2 PPR1 RAP1 RCS1 RCS1 RDR1 RDS1 RDS2 RDS3 SFP1 SFP1 SKN7 SKN7 SMP1 SOK2 STB5 STE12 STP1 SUT1 SWI4 TBF1 TOS4 TOS8 TYE7 TYE7 UGA3 WAR1 XBP1 YAP1 YAP1 YAP3 ZMS1 Factors:6 (how much variance is explained) Correlations: ADR1 X Factor 1: (P=0.0008) ADR1 X Factor 2: (P<0.0001) ADR1 X Factor ? – non significant. Transcripts to explore or confirm.
Exploratory Factor Analysis ADR1 Transcription Factor Which of the regulated genes are real? Loading Gene:Factor 1Factor 2 ABF AFT2-454 * CHA4-51 CRZ1-754* FZF150 *10 GAT254*64* GAT346*-1 GAT465*-21
Confirmatory Factor Analysis, Structural Equation 31 Exogenous variables 12 Endogenous parameters
Yeast Pathways Are “Well”-Established V1 = 0.54V2 + E1 t = V2 = 0.61V3 + E2 t = 9.69 V3 = 6.60V4 – 2.57V5 +E3 t = 4.19 t=1.95 model phenotype; Direct selection on V1 indirect selection.
Yeast Confirmatory Factor Analysis Genetic / metabolic Small natural mutations introduce are networks non-linear;linear deviations (natural variation is mostly additive); To parameterize the Nature supplies unlimited number of model, many degrees ofsegregating alleles; freedom are required; Pathways are never Latent variables can be used complete, many variables instead. can not be measured.
33 Exogenous variables 13 Endogenous parameters Yeast Confirmatory Factor Analysis, Latent Variables
1. Intraspecific Transcriptome Variation 10-20% genes significantly vary in transcript level among populations, heritability is high; there is a similar level of variation for alternative splicing levels; transcriptome variation is profoundly: heritable in males, epistatic in females; antagonistic effects of alleles on two sexes may contribute to the overabundance of X-linked variation; factors explaining ITV is a promising analysis to identify genes and networks controlling QTs.