plotProfile (deepTools) generated a peak on gene body region

Question

[mbk0asis]

Hi, all!

I'm trying to plot the DNA methylation profiles using deepTools and having annoying peaks on the gene body region as you can see the picture below.

The code I used;

$ computeMatrix scale-regions -S Sample1.bw Sample2.bw Sample3.bw -R genes.bed -a 5000 -b 5000 -o Samples.matrix --regionBodyLength 15000 --skipZeros
$ plotProfile -m Samples.matrix -out Samples.profile.png --perGroup

How can I remove them?

Thank you!

[Shaurya Jauhari]

Hello.

Not sure if my problem aligns perfectly to this thread, but I want to share some weird plots from the computeMatrix-plotProfile workflow. Not sure about "flatness" of the plot.

computeMatrix scale-regions \
-R LD9783.bed \
-S H3K4me1.bw \
H3K4me2.bw \
H3K4me3.bw \
H3K27ac.bw \
-p 1 \
-—regionBodyLength 2000 \
-a 1000 -b 1000 \
--samplesLabel "H3K4me1" "H3K4me2" "H3K4me3" "H3K27ac" \
-o A549_LD9783.gz \
--outFileNameMatrix A549_LD9783.tab \
--outFileSortedRegions A549_LD9783.bed

The corresponding plotProfile script is as under.

plotProfile -m A549_LD9783.gz -out A549_LD9783.png --numPlotsPerRow 2 --plotTitle "A549|LD9783"

The plot is very boxy and abrupt. I have checked the bedGraph values and they aren't uniform as such.

Any suggestions on what could be erroneous here.

[GenoMax]

Use the directions: How to add images to a Biostars post

Create a new thread if this is not directly related to the original question.

Answer 1

[Devon Ryan]

Give plotHeatmap a try, I expect you have some outlier bin in one gene (just remove that gene, though in the future you'll want to blacklist that region when making bigWig files) that will be apparent in the heatmap.

[mbk0asis]

Thank you. I will give it a try.

[mbk0asis]

I am stuck at the first step which is ID the outliers. I was going to export the matrix using "–outFileNameMatrix" option and edit the matrix. Then, how can I make the edited file back to the matrix file 'plotHeamap' recognizes? Should I just use other tools like R to draw the heatmap?

[Devon Ryan]

If you want to actually edit the heatmap then don't bother with --outFileNameMatrix, just gunzip the output from computeMatrix, remove the problematic line, change the group_boundaries so they limits are correct, and then gzip the results back up.

In general, it's probably simpler to just figure out where the problem region is and remake the bigWig files with that region blacklisted.

[mbk0asis]

Your second suggestion sounds much easier, haha. Thank you!