Filter BINDetect results based on volcano plot

[KeyTals]

Hello,

Thanks you for making very useful tool, I have question regarding excel and text file crated from BINDetect step, Is it possible to filter the "diff_results" files with only entries of TFs/motifs which are presented as up any condition from volcano plot ( which are colored red or green)?

I am aware of critetia of " -log10(p-value) above the 95% quantile and/or differential binding scores smaller/larger than the 5% and 95% quantiles (top 5% in each direction)" but still diff_results file list for individual motifs are very big then the motifs/TFs identified as up from volcano plot.

Thanks,
Keyur

[msbentsen]

Hi Keyur,

Sorry no, there is no direct way other than maybe writing a small script to do so for each contrast. The information is located in the "bindetect_results.txt"-file.

If you are familiar with Python, you might use Pandas to read the text into a dataframe and then setup the criteria for each contrast. This will give you a "colored : yes/no" per TF per contrast. Then, based on the criteria, you can choose to only include TFs up/down in any of the contrasts. The change and pvalue columns are always named as <condition1>_<condition2>_change and <condition1>_<condition2>_pvalue, so it should not be too difficult to access the column data.

I am sorry I can't be of more help, and I hope you find a solution.

Best,
Mette

[DittmanC]

is there any way to adjust the cut-off of the l2fc? I have been trying the --pesudo option, but it doesn't change

[msbentsen]

is there any way to adjust the cut-off of the l2fc? I have been trying the --pesudo option, but it doesn't change

Unfortunately not, no. The --pseudo option only changes the pseudocount added before log2fc-calculation, but not the threshold for up/down-regulation. However, you might have a look within the bindetect_results.txt-file and set your own threshold from the <condition1>_<condition2>_change column. That way, you get a list of TFs that are either more/less strict depending on your application.

[Adrien-Evo]

Here is my code in R to change how the volcanoPlots looks:

library(ggplot2)
library(biomaRt)
library(ggrepel)

changelimit = 0.09
pvallimit = 100

topgenes = read.table(bindetect_results.txt, stringsAsFactor = FALSE, h=TRUE)
topgenes$diffexpressed <- rep("NO",length(topgenes$WT_KI_change))

topgenes$diffexpressed[topgenes$WT_KI_change > changelimit & -log10(topgenes$WT_KI_pvalue) > pvallimit] <- "UP"
topgenes$diffexpressed[topgenes$WT_KI_change < -changelimit & -log10(topgenes$WT_KI_pvalue) > pvallimit] <- "DOWN"

topgenes$labels <- rep(NA,length(topgenes$WT_KI_change))
topgenes$labels[topgenes$diffexpressed != "NO"] <- topgenes$name[topgenes$diffexpressed != "NO"]


topgenes$diffexpressed <- as.factor(topgenes$diffexpressed)

p <-ggplot(data=topgenes, aes(x=WT_KI_change, y=-log10(WT_KI_pvalue),label=labels)) + 
    geom_point(aes(colour = topgenes$diffexpressed))+ 
    theme_classic() + 
    geom_text_repel(aes(colour = diffexpressed),max.overlaps = Inf,show.legend = FALSE,size = 5) + 
    scale_colour_manual(values=c("red","grey","blue"),labels=c("Higher scores in CONE1", "No changes", "Higher scores in COND2")) +
    geom_vline(xintercept=c(-changelimit, changelimit),linetype = "dashed",  alpha=0.4) +
    geom_hline(yintercept=pvallimit, linetype = "dashed", alpha=0.4)  +
    theme(legend.position = c(0.8,0.2),legend.key.size = unit(0.4, "cm"),legend.text=element_text(size=20),axis.text.y = element_text(size = 15),axis.text.x = element_text(size = 15),axis.title=element_text(size=18)) + 
    labs(x ="Change", y = expression("Significance (-Log[10])")) + 
    expand_limits(x=c(-0.2,0.2)) + 
    guides(colour=guide_legend(title=NULL))

Just change the WT_KI_change to the name of your column