Question: ChiP Seq Track height and scaling in UCSC Browser
gravatar for Bao Ho
22 months ago by
Bao Ho80
Bao Ho80 wrote:

Hi all,

I recently work on several ChIP-seq data sets. My traditional workflow is .fastq/.fastq.gz -> .sam (via Bowtie) -> MACS (after BAM conversion) -> wig to bigwig conversion. Thus, my usual output are .bw track files that I then view through UCSC genome browser, with the y axis (track height) scaled to sequencing depth (for example the Bowtie alignment reported 15M aligned reads, y axis would be 1 - 15).

However, recently I encountered an issue with MACS, or more precisely, wigToBigWig tool:

Overlap on chr4 between items starting at 29999999 and 30000000. Please remove overlaps and try again Fatal error: Matched on Error Error running wigToBigWig.

I followed the suggestion of the Galaxy team and tested out MACS 2 and succeeded constructing a BedGraph file, which I then converted to .bw track to view through UCSC genome browser as usual. The result was very good. However, I noticed that the y-axis scaling is different from what I usually have with MACS -> bigwig method. In MACS2 -> bigwig case the peaks are only visible with y-axis range being 0.4 - 0.6, while in traditional MACS -> bigwig case peaks are usually visible with the range being 1 - 15 (as I mentioned above). Is there a way to change the scaling of the track created by MACS2 -> bigwig method to match the scaling scheme that I usually use for with MACS -> bigwig?

Here is the link to the history in question:

I must admit that my understanding of ChIP-seq analysis is very shallow. Please feel free to critique my practices. I would greatly appreciate an explanation for the meaning of track height if anyone can provide it.



ucsc viz bigwig macs macs2 • 1.2k views
ADD COMMENTlink modified 21 months ago by Jennifer Hillman Jackson25k • written 22 months ago by Bao Ho80

Track visualization can be customized at UCSC (see their help docs, or just click into the track details and adjust).

Conversion using wigToBigWig is not the source of the score/depth difference - the underlying statistics come from MACS/2 output. My initial guess is that there is a difference in how the parameters used with the MACS vs MACS2 runs were applied that impacted scores/scaling. There are other differences between the two tool versions that will make any rerun job using the newer tool version produce slightly different results, even when compatible parameters are used.

I'll take a look at the new runs (and others can as well!). Thanks for testing out this suggested solution to the prior problem. More feedback after review. - Jen

ADD REPLYlink written 22 months ago by Jennifer Hillman Jackson25k

I have tried to run MACS and MACS2 on the same alignment result for several data sets. Surprisingly the results I received are very similar. In some cases MACS2 yielded a higher level of noise/background. It seems that the data that I shared above is the only unique case with tiny peaks and clean background.

I have also found this:

According to the website the scoring schemes for MACS and MACS2 are the same (except quality score, but I do not think that this affect the height of peaks).

Thank you for your help! If you have found something of note please let me know.

ADD REPLYlink written 22 months ago by Bao Ho80
gravatar for Jennifer Hillman Jackson
21 months ago by
United States
Jennifer Hillman Jackson25k wrote:


I found the difference - the "Effective Genome Size" differs significantly between the MACS and MACS2 jobs in the shared history. I suspect that if the same value was used, the results would align as they did in your other test runs with different samples.

The "tag" size is also (very) slightly different. With MACS, this is user specified and 50 was entered. With MACS2, this is interpreted by the tool from the fastq input and the tag size is actually 51. Run FastQC when you wish to learn real tag size (fastq sequence length) when using MACS.

I didn't look further into differing parameters, but if you want to pursue this more, that is what I suggest reviewing.

Hope this helps and that you can move forward with your analysis now! Jen, Galaxy team

ADD COMMENTlink modified 21 months ago • written 21 months ago by Jennifer Hillman Jackson25k

Thank you! I will try to make output from MACS and MACS2 consistent & similar. I will report my further findings later.

ADD REPLYlink written 21 months ago by Bao Ho80
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 171 users visited in the last hour