kinase cardiotoxicity v2

chip <- read_excel("ChIPseq Trametinib mouse hearts.xlsx");
mib <- read_excel("MIB.xlsx"); #every value is sig for at least one rep
kin <- read.csv("kinase.csv"); #every value here is p <.05
tram <- read.csv("tramet.csv");

#every p val for tram that is p<.05
tram <- tram %>%
    filter(pval < .05) %>%
    filter(!Mean.1 == 0);

1. 116 significant genes from at least one rep in the MIB data
2. 5183 significant genes from the trametinib data

overlap <- merge(x=tram, y=mib, by.x=c("Gene"),
    by.y=c("Gene...2"));

overlap <- overlap[,c(1,13,49,51,52,54,55,57)]

#add column of average MIB values
overlap$avg <-apply(overlap[,c(3,5,7)],1,mean);

1. There are 43 overlapping genes of significance between MIB and tram data, inclusive of all reps of MIB.

#filter out the MIB reps that were not sig
df <- overlap %>%
  filter(!overlap$`Significant?...38`=="no" & !overlap$`Significant?...44`=="no");

#make the rna seq pval numeric
df$pval <- as.numeric(df$pval)

1. When non-significant MIB reps are removed, there are 26 overlapping significant genes.

Looking at Trametinib RNAseq p-values

#ALL TRAM GENES THAT ARE P<.05
p <- tram %>%
  ggplot( aes(x=as.numeric(pval))) +
    geom_histogram( color="#e9ecef", alpha=0.6, position = 'identity', bins=40, fill = "#404080") +
    #scale_x_continuous(breaks = round(seq(min(df$pval), max(df$pval), by = .01),1)) +
    labs(fill="") + gghisto

#ONLY TRAM GENES THAT OVERLAP WITH MIB
 p0 <- df %>%
  ggplot( aes(x=pval)) +
    geom_histogram( color="#e9ecef", alpha=0.6, position = 'identity', bins=25, fill = "#404080") +
    #scale_x_continuous(breaks = round(seq(min(df$pval), max(df$pval), by = .01),1)) +
    labs(fill="") + gghisto

 grid.arrange(p,p0)

Looking at tram MIB avg values

#for the genes that overlap at least ONE mib rep and the tram RNAseq
p1 <- overlap %>%
  ggplot( aes(x=avg)) +
    geom_histogram( color="#e9ecef", alpha=0.6, position = 'identity', bins=22, fill = "#404080") +
    scale_x_continuous(breaks = round(seq(-6, 2, by = .2),1)) +
    labs(fill="") + ggtitle("Distribution of avg MIB values for at least 1 rep overlapping gene") + gghisto

#for the n=26 genes that overlap RNAseq and all three reps of the MIB data
p2 <- df %>%
  ggplot( aes(x=avg)) +
    geom_histogram( color="#e9ecef", alpha=0.6, position = 'identity', bins=22, fill = "#404080") +
    #scale_x_continuous(breaks = round(seq(min(overlap$avg), max(mib$AVG...36), by = .4),1)) +
    labs(fill="") + ggtitle("Distribution of avg MIB values for ALL reps overlapping genes") + gghisto

grid.arrange(p1,p2)

p3 <- ggplot(df, aes(pval, avg, label=Gene)) + geom_point(aes(color="pink",size=2,alpha=.7)) + ggtitle("RNAseq pval vs MIB avg") + gghisto + theme(legend.position = "none") + 
  geom_label_repel(aes(label = Gene),
                  box.padding   = 0.25, 
                  point.padding = 0.4,
                  segment.color = 'grey50')

p4 <- ggplot(df, aes(avg, pval, label=Gene)) + geom_point(aes(color="pink",size=4,alpha=.7)) + ggtitle("RNAseq pval vs MIB avg") + gghisto + theme(legend.position = "none") + 
  geom_label_repel(aes(label = Gene),
                  box.padding   = 0.25, 
                  point.padding = 0.4,
                  segment.color = 'grey50')

The variable pval gives the p-value from the trametinib RNAseq analysis. The variable avg takes the average value from the three replicates of the MIB data. Here, these two valyes are correlated.

#grid.arrange(p1,p2,nrow=1)
p3;p4

correlation

cor(df$pval,df$avg)

[1] -0.03682366

There appears to be no relationship between protein average and RNAseq p-value.