Q: MACS2 parameter for ATAC-seq, BAMPE or single-end mode

[yue131]

Hi, I am tring to call peaks using macs2 from ATAC-seq data.

many papers use --shift -100 --extsize 200 for MACS2 rather than -f BAMPE; but my data is paired_end reads，if I use -f BAMPE, then the parameter of --shift -100 and --extsize 200 will be discard.

so, I have some questions:
if I am interested in genome_wide chromatin accessibility, which parameter is better?

[taoliu]

Hi @yue131 , I would recommend using the paired-end mode through -f BAMPE if you are focusing on the accessible region between the Tn5 insertion sites of a read pair. John Gaspar discussed this in his biorxiv paper: https://www.biorxiv.org/content/10.1101/496521v1. And his modifications have been introduced to MACS2 since v2.1.2.

Also, you may want to try HMMRATAC (https://github.com/LiuLabUB/HMMRATAC) refer to #145.

[yue131]

thank you for your reply.

[taoliu]

You also found the discussion in #195, where @kwcurrin had a good point that in certain circumstances, where you have a dominant abundance of the nucleosomal fragments in the ATAC-seq library, the single-end mode is more reasonable. However, a not-under-digested ATAC-seq library should show a fragment distribution with more abundant 'short fragments' than 'nucleosomal fragments.' Because ATAC-seq is mainly used to detect the regulatory elements instead of nucleosome positioning, and you want more Tn5 insertion sites in TF binding sites comparing to nucleosome-linker regions. If so, utilizing the fragment length information from your PE library should lead to a more precise enrichment at the accessible chromatin. Anyhow, as we discussed in the HMMRATAC paper (https://academic.oup.com/nar/article/47/16/e91/5519166), more sophisticated modeling on the fragment distribution will give better precision and recall.

[yue131]

this is my fragment length distribution. approximately ½ of reads are of sub-nucleosomal length (less than approximately 150 bp) and approximately half of the reads longer than this length. so in this circumstances, which parameter is better? -f BAMPF or --shift -100 --extsize 200?

[kwcurrin]

Hello,

Unless you want to use a peak caller specific to ATAC-seq, I would recommend using --shift -100 --extsize 200 because it is preferrable for fragments that span nucleosomes and is a decent approximation for fragments less 150 bp. Even though the <150bp fragments are often the largest hump in the fragment length distribution, they aren't necessarily the majority of fragments. Your case is a good example of this.

---

From: yue131 [notifications@github.com] Sent: Saturday, October 19, 2019 12:03 PM To: taoliu/MACS Cc: Currin, Kevin; Mention Subject: Re: [taoliu/MACS] Q: MACS2 parameter for ATAC-seq, BAMPE or single-end mode (#331)

[image]https://user-images.githubusercontent.com/56757131/67147815-179da100-f2cb-11e9-96e0-90f5074573f4.png this is my fragment length distribution. approximately ½ of reads are of sub-nucleosomal length (less than approximately 150 bp) and approximately half of the reads longer than this length. so in this circumstances, which parameter is better? -f BAMPF or --shift -100 --extsize 200?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/taoliu/MACS/issues/331?email_source=notifications&email_token=AC7PRTEP5LON2MO3IKD6FOTQPMVVTA5CNFSM4JCPFPBKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBXU2XQ#issuecomment-544165214, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AC7PRTECJ4XDXMMELHAL4L3QPMVVTANCNFSM4JCPFPBA.

[yirenheihei]

hello,using -f BAMPE,can you get your results? now i also analyze ATAC-seq,but i dont know which parameter is ok?thank you

[yue131]

yes。

[yue131]

@yirenheihei yes! you can -f BAMPE

[yirenheihei]

thanks @yue131