sliding windows with or without overlap?

Question

[Yu]

Dear all,

I am confused about sliding window method, we often using this method to analysis ChIP-seq data or Copy number variation(CNV).

Sliding windows w/o overlap is divided genome into several equal segments, it often used in CNV analysis. Sliding windows w/ overlap is like moving average in statistics.

I read the thread Question: Sliding Window for CNV Analysis, but it doesn't compare those two methods. Is sliding windows with overlap better than sliding window without overlap? Should we using sliding window with overlap on ChIP-seq data analysis?

Answer 1

[Ibrahim Tanyalcin]

Dear,

I will make an educated guess. Consider the series 1-3-4-6-15-1-0-9-(-1)-5-10-10

Now if I were to sum each 2 consecutive number, take average and report it back with no overlap it would be:

2-2

----5-5

-------- 8-8

-------------4.5-4.5

----------------------2-2

--------------------------10-10

In total I did 6 average operations.

However if I were to sum 1st and 2nd, then 2nd and 3rd and so on:

2-(2 or 3.5)

---------------(3.5 or 5)-(5 or 10.5)

------------------------------------------(10.5 or 8)-(8 or 0.5)

---------------------------------------------------------------------(0.5 or 4.5)-(4.5 or 4)

--------------------------------------------------------------------------------------------------(4 or 2)-(2 or 7.5)

-------------------------------------------------------------------------------------------------------------------------(7.5 or 10)-(10 or 10)

In total I did 12 average operations. As a result overlaping method should take more computation time but it should provide more data/resolution. You will be able to see more between the points and so on.

[Yu]

Thank you. Sliding windows may provide more resolution.