search

ZW-xjtlu / exomePeak2

Peak calling and differential methylation for MeRIP-Seq
25 stars 5 forks source link

How to decide the fragment_length #3

Open sunhaifeng123 opened 4 years ago

sunhaifeng123 commented 4 years ago

Hi,

Recently I have a question about how to decide my fragment_length:

For example my fragment length peak is 200 (not including adapters), and I take PE150 for sequence, should I change the default fragment_length=100 to 200 ? And how about taking SE50 for sequence?

This really puzzled me these days.

Thanks advance for your reply!

Haifeng NanJing Medical University 2020-08-04

ZW-xjtlu commented 4 years ago

Hi,

Recently I have a question about how to decide my fragment_length:

For example my fragment length peak is 200 (not including adapters), and I take PE150 for sequence, should I change the default fragment_length=100 to 200 ? And how about taking SE50 for sequence?

This really puzzled me these days.

Thanks advance for your reply!

Haifeng NanJing Medical University 2020-08-04

Hi Haifeng,

Please set the fragment_length parameter to the average fragment lengths of your libraries (excluding the adapter length).

For PE data, we can directly estimate the average fragment length from the alignment results and transcript annotations. However, due to the large variation in fragment length within samples and datasets, the estimations on individual samples are computationally expensive and may introduce technical instabilities, so it has not been implemented in exomePeak2.

If you can determine that the average fragment length of the PE150 library is 200 across all of your IP and input samples, then just set this parameter to 200. For SE50, you can just use the default value of 100, because SE reads usually have shorter fragment lengths than PE reads.

It is fine if the fragment_length parameter is shorter than the actual fragment length, but if it is much longer than the actual fragment length, the performance of peak calling will be impaired.

If you have any other questions, you are welcome to ask me later in this issue.

Best wishes, Zhen Xi'an Jiaotong-Liverpool University

sunhaifeng123 commented 4 years ago

Hi Zhen,

Thanks very much for your patient answers !

I will try to adjust the fragment_length parameter.

Best wishes, Haifeng