Post date: Dec 24, 2014 10:50:59 PM
I tested for a correlation between observed het. (prop. of het. loci) and survival for the A and C treatments (R code below). I used the integer genotypes, so that het = 1. Het. had no affect on C, but a slight (significant) effect on A. This makes some sense as the source was A and increased het. could mean recent migrant ancestry.
nl<-8151262
ni<-238
GA<-matrix(scan("G_int_timemaA.txt",n=nl*ni,sep=","),nrow=nl,ncol=ni,byrow=T)
hetA<-apply(GA==1,2,mean)
pA<-read.table("pheno_samplesA.txt",header=F)
out<-glm(as.numeric(pA[,1]) ~ hetA, family=binomial)
summary(out)
#Deviance Residuals:
# Min 1Q Median 3Q Max
#-1.0827 -0.8423 -0.7454 1.4501 2.1893
#Coefficients:
# Estimate Std. Error z value Pr(>|z|)
#(Intercept) 9.737 5.175 1.881 0.0599 .
#hetA -103.244 50.038 -2.063 0.0391 *
#---
#Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#(Dispersion parameter for binomial family taken to be 1)
# Null deviance: 281.03 on 237 degrees of freedom
#Residual deviance: 276.58 on 236 degrees of freedom
#AIC: 280.58
#Number of Fisher Scoring iterations: 4
tapply(X=hetA,INDEX=pA[,1],mean)
# 0 1
#0.1040449 0.1031265
nl<-8151262
ni<-235
GC<-matrix(scan("G_int_timemaC.txt",n=nl*ni,sep=","),nrow=nl,ncol=ni,byrow=T)
hetC<-apply(GC==1,2,mean)
pC<-read.table("pheno_samplesC.txt",header=F)
out<-glm(as.numeric(pC[,1]) ~ hetC, family=binomial)
summary(out)
#Call:
#glm(formula = as.numeric(pC[, 1]) ~ hetC, family = binomial)
#Deviance Residuals:
# Min 1Q Median 3Q Max
#-0.8229 -0.8199 -0.8188 1.5813 1.5902
#Coefficients:
Estimate Std. Error z value Pr(>|z|)
#(Intercept) -0.7566 4.8342 -0.157 0.876
#hetC -1.5675 46.5652 -0.034 0.973
#(Dispersion parameter for binomial family taken to be 1)
# Null deviance: 280.92 on 234 degrees of freedom
#Residual deviance: 280.92 on 233 degrees of freedom
#AIC: 284.92
#Number of Fisher Scoring iterations: 4
Interestingly, there is a marginally significant negative effect of het on stripe on A, but nothing on C. This (at least the first part) makes sense because A favors stripe, thus immigrants with higher het should be less likely to have stripes.
Call:
glm(formula = as.numeric(sA[, 1]) ~ hetA, family = binomial)
Deviance Residuals:
Min 1Q Median 3Q Max
-1.8705 0.3311 0.6374 0.7660 1.0456
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 18.14 10.02 1.809 0.0704 .
hetA -164.18 96.52 -1.701 0.0890 .
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 73.109 on 65 degrees of freedom
Residual deviance: 69.804 on 64 degrees of freedom
(172 observations deleted due to missingness)
AIC: 73.804