Chip-Seq: best value/money read length SE/PE ?

[eslondon]

Hello there,

I have asked this to several collaborators but always got quite a varied response. If you are designing a new experiment from scratch and want the most reasonable quality/money ratio, what read length would you use for Chip-Seq and would you use paired-end by all means over single-end?

I have mostly experience with RNA-Seq where it is clear that using longer reads (e.g. 75bp) and Paired-End reads gives much richer information on splicing. (We avoid actively the new 100bp reads in part because it's not very good value for money and in part because it is too new). But what about Chip-Seq?

Several papers have published so far good results with SE short reads. On the other hand longer reads and paired-end reads should allow improved identification of tougher regions, e.g. with repeats, as well as more accurate resolution of the peak (from both directions).

Comments?

thanks!

Elia

[Xi Wang]

Hi, Elia,

I am a bioinformatics person. I suggest you use long read length and PE reads. The reasons are: 1. long read length will reduce the multi-alignment reads, which are due to some simple sequence repeats in the genome, 2. PE reads will also reduce the multi-reads, 3. an important parameter to locate binding sites is the length of DNA fragments, if PE is used, the length of fragments are determined through mapping, rather than estimated from the read profile peaks between the two strand. A possible byproduct will be some new methods for dealing with Chip-seq data with longer reads and PE reads.

Xi