macs14 replicates treatment and final wiggle peak files

Question

[kashiff007]

Hi,

I have few quick question regarding macs14?

• How one can treat the replicates? Do I have to run separately and then take average from the final peaks or is there any way to include it in one run?

• macs14 generates two wig files under "--single-profile" option; one for control and one for treated. which file to use for plotting final peak profile? Do I have to subtract treated - control manually or does it give the final peak wig files?

Thanks in advance

Answer 1

[GouthamAtla]

If you have the replicate, you can take the consistent peaks called in both replicates. There will be variability in ChIP-Seq data, so sometimes you may want to consider union of peaks instead of intersections.

Regarding peak profile, by control do you mean input ? or control in the sense of WT/KO ? You can plot both the treated and control using deeptools2 so that it will be clear that both separated if that's the intention. For UCSC tracks, I would get the bigwigs with -log10(p-value), to show enrichment over inputs.

[kashiff007]

By control I mean input here. Can I use treated wig/bw directly to see my enrichment of do I have to do manual subtraction of treated wig - control (input) wig file?

"I would get the bigwigs with -log10(p-value), to show enrichment over inputs." will this file be generated or do I have to use some other file to generate this?

I am asking this because my folder NAME_MACS_wiggle has two separate folders naming "control" and "treated" and each of them having one wig file; NAME_control_afterfiting_all.wig.gz, NAME_treat_afterfiting_all.wig.gz respectively.

[GouthamAtla]

By control I mean input here. Can I use treated wig/bw directly to see my enrichment of do I have to do manual subtraction of treated wig - control (input) wig file?

You can use your treated file directly but its better to show the input as well so that it will be more convincing. Are you talking about ucsc tracks or a profile plot over promoters/enhancers ?

"I would get the bigwigs with -log10(p-value), to show enrichment over inputs." will this file be generated or do I have to use some other file to generate this?

You can run MACS2 sub command like this:

macs2 bdgcmp -t treat.pileup.bdg -c local_lambda.bdg -m ppois -o pvalue.bdg which contains -log10(p-values) for each base pair which can be converted to bigwig (bedGraph to bigWig)

I am asking this because my folder NAME_MACS_wiggle has two separate folders naming "control" and "treated" and each of them having one wig file; NAME_control_afterfiting_all.wig.gz, NAME_treat_afterfiting_all.wig.gz respectively.

This is because its good to show both treat and input so that its more convincing that your peaks are real.

[kashiff007]

Thanks @geek_y for such detail explanation. Al thought have you ever encounter the following error while running the macs14. "error occurred: Invalid variableStep wig file format in line starting with "variableStep chrom=chr3 span=10"

what could be the reason for this?

Thanks

[GouthamAtla]

MACS doesn't work with bigwig. It works with bed graph files.