Analyze Tie1F2.Rmd

## Tie1 F2s

Quick analysis to determine if we should look at more F2s to distinguigh between GA2ox and the Alpha Glucosidase as causative for tie-1 phenotype

Get the data and take a quick look
```{r}
tie1 <- read.csv("tie.F2s.csv",na.strings=c("NA","-"))
summary(tie1)
head(tie1)
tie1$gt <- paste("GA2ox",tie1$GA2,"_AG.2",tie1$AG,sep="")
tie1$height <- tie1$epi + tie1$int1 + tie1$int2
tie1$int.total <- tie1$int1 + tie1$int2
```

Remove unused columns
```{r}
tie1.small <- tie1[,!grepl("X",colnames(tie1))]
#tie1.small <- tie1.small[tie1.small$first.leaf=="1",] #remove wierd plants where there are 2 leave at first node.
head(tie1.small)
```

melt
```{r}
library(reshape2)
tie1.melt <- melt(tie1.small,id.vars = c("plant","GA2","AG","gt","flat","first.leaf"),variable.name="organ")
head(tie1.melt)
```

plot it
```{r}
library(ggplot2)
pl <- ggplot(na.omit(tie1.melt),aes(x=gt,y=value))
pl <- pl + geom_boxplot()
pl <- pl + facet_wrap( ~ organ)
pl <- pl + theme(axis.text.x  = element_text(angle=90, vjust=0.5))
pl + geom_point()
```

## test it for epicotyl

### GA2oxidase only
```{r}
lm1 <- lm(epi ~ GA2,data=tie1)
summary(lm1)
```
Very strong effect and we can see the het effect also.

### Alpha-glucosidase only
```{r}
lm2 <- lm(epi ~ AG,data=tie1)
summary(lm2)
```
Yes there is an effect here but probably due to linkage with GA2ox.  Test via additive model.

### Additive model
```{r}
lm3 <- lm(epi ~ GA2 + AG,data=tie1)
summary(lm3)
```
This confirms that when both loci are in the model that only GA2 genotype matters 

### interaction model
To test whether or not there is some wierd interaction between loci
```{r}
lm4 <- lm(epi ~ GA2*AG,data=tie1)
summary(lm4)
```
Still no evidence for alpha-glucosidase playing a role.

### an alternative
Predict on two-locus genotype
```{r}
lm5 <- lm(epi ~ gt,data=na.omit(tie1))
summary(lm5)
```



## test for total height

### GA2oxidase only
```{r}
lm1 <- lm(height ~ GA2,data=tie1)
summary(lm1)
```
Very strong effect and we can see het effect also.

### Alpha-glucosidase only
```{r}
lm2 <- lm(height ~ AG,data=tie1)
summary(lm2)
```
Yes there is an effect here but probably due to linkage with GA2ox.  Test via additive model.

### Additive model
```{r}
lm3 <- lm(height ~ GA2 + AG,data=tie1)
summary(lm3)
```
This confirms that when both loci are in the model that only GA2 genotype matters 

### interaction model
To test whether or not there is some wierd interaction between loci
```{r}
lm4 <- lm(height ~ GA2*AG,data=tie1)
summary(lm4)
```

### an alternative
Predicted on two-locus genotype
```{r}
lm5 <- lm(height ~ gt,data=na.omit(tie1))
summary(lm5)
```



## test for total internode

### GA2oxidase only
```{r}
lm1 <- lm(int.total ~ GA2,data=tie1)
summary(lm1)
```
Very strong effect and we can see het effect also.

### Alpha-glucosidase only
```{r}
lm2 <- lm(int.total ~ AG,data=tie1)
summary(lm2)
```
Yes there is an effect here but probably due to linkage with GA2ox.  Test via additive model.

### Additive model
```{r}
lm3 <- lm(int.total ~ GA2 + AG,data=tie1)
summary(lm3)
```
This confirms that when both loci are in the model that only GA2 genotype matters 

### interaction model
To test whether or not there is some wierd interaction between loci
```{r}
lm4 <- lm(int.total ~ GA2*AG,data=tie1)
summary(lm4)
```
This argues that the alpha-glucosidase being het instead of Heinz decreases height in the presence of GA
### an alternative
Predicted on two-locus genotype
```{r}
lm5 <- lm(int.total ~ gt,data=na.omit(tie1))
summary(lm5)
```

### make a figure

```{r}
#make a function for summarizing
mean.sem <- function(x,na.rm=TRUE) {
  if(na.rm) x <- na.omit(x)
  mean.x <- mean(x)
  sem.x <- sd(x)/sqrt(length(x))
  data.frame(
    y=mean.x,
    ymin=mean.x-sem.x,
    ymax=mean.x+sem.x)
}

#plot AG conditioned on GA2
pl <- ggplot(data=na.omit(tie1),aes(x=AG,y=int.total))
pl <- pl + geom_bar(stat="summary",fun.data="mean.sem",fill="skyblue")
pl <- pl + geom_errorbar(stat="summary",fun.data="mean.sem",width=0.5)
pl <- pl + facet_wrap( ~ GA2,nrow=1)
pl + ylab("Total Internode Length") + xlab("Alpha-glucosidase Genotype")
ggsave("AG2_total_internode.pdf",height=6,width=6)

#plot GA2 conditions on AG
pl <- ggplot(data=na.omit(tie1),aes(x=GA2,y=int.total))
pl <- pl + geom_bar(stat="summary",fun.data="mean.sem",fill="seagreen")
pl <- pl + geom_errorbar(stat="summary",fun.data="mean.sem",width=0.5)
pl <- pl + facet_wrap( ~ AG,nrow=1)
pl + ylab("Total Internode Length") + xlab("GA2 Oxidase 7/8 Genotype") 
ggsave("GA2_total_internode.pdf",height=6,width=6)
```