Hello everyone,
Could some please explain to me what do the data values in a WIG file mean? I have read the docmentaion about this format on the UCSC webpage and also checked for an answer on the biostar forum.
This question is a bit similar to the one posted earlier:
https://www.biostars.org/p/16261/
I understand that a WIG file can have data values as log2(IP/input), p-values, base coverage, etc. but I am trying to understand this in terms of peaks called using MACS.
On performing peak calling using MACS version 1.4.2, the header for the wiggle file says "Extended tag pileup from MACS version 1.4.2 20120305 for every 10 bp".
Does this mean that the data value in the WIG file at a given location represents the number of reads aligned within a 10bp window at that location? To check this, I visualised the WIG file along with its BAMs (Input and ChIP) on IGV but didnt see a coherence between the two. For instance, one peak had 18 reads at its summit while the data value in the WIG here was 63 and the peak constituted of a total of 94 reads. The read length is 50 bp while d= 211 (MACS) and so I checked for the total number of reads (and also number of reads extended reads d or d/2) within a window of length 'd' or 'd/2' centered at the summit (WIG value 63) and even then the read count was not 63. In addition, for regions without any aligned read (outside the peak region) the data value was 1.
So how does MACS calculate the data value for WIG and what does it mean if not the read count?
EDIT: Duplicate reads from the BAM file were removed prior to peak calling and hence duplicate-free BAM was compared with its wiggle file.
https://www.biostars.org/p/16261/
This link may help:
https://groups.google.com/d/msg/macs-announcement/LZpliDkdN-8/sO0uIr8fXdkJ
Hey Matej,
Thanks for the reply. I forgot to mention in my question that all duplicates from the BAMs were removed before peak calling and I compared this duplicate removed BAM with its peak called wiggle file. So, the coverage and extended read count valueat any particular position should be the same.
Right?
Dear Rishikesh
It has been a while I have worked on MACS or MACS2, but as far as I can recall, the counts you are looking at using IGV and using MACS are different because of the Read Shifting that MACS will perform before finding the centre of the peak - you have to refer to their main paper to read the details of how the shifting works. As IGV does not perform that, it may not be concordant. Furthermore, you may need to count the reads from both strands, i.e. plus and minus.
Cheers
Hey Umar,
Thanks for your reply. After the link sent by Matej I rechecked the peaks and yes I figured out that was the very thing I was not taking into consideration. I checked the BAM files and extended reads from both strands at a peak location and then the extended read count matched with the WIG values. Thanks again to the both of you.
No problem, glad to be of help. As an additional point, if it is ChIP Seq data, then it does not matter, but for MeDIP Seq, due to the way the library is prepared, you can separate the strands from where the signal comes from, by performing an additional step in the analysis. https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-016-1158-7
Hi Benxia,
You can extend the reads through IGV. For this, your IGV installation needs to be version 2.3.50 or later.
To extend the reads, go to:
Open IGV>Tools> Run IGV tools.
Make sure the command at the top is selected to 'count'. Your input file will be a BAM file and the output is default .tdf. If you wish you can change this to .wig
Under the TDF and count options you will find a field for 'Extension factor'. Enter the desired length of extension. E.g. If your reads are 50bp long and if your fragment size is 300bp, then enter 250 in the 'extension factor' field.
Click on run.
Once complete you can load this .tdf or .wig file in the IGV and view your results.
Hi Benxia,
I have not tried extending reads with any code but only tested IGV for my problem. Probably the thread below might help.
https://www.biostars.org/p/49163/
I have not tried it out but maybe it might help.
- Badges
- Science method