Very different result from chip-sea data between mm9 and mm10 using galaxy. Is that normal?

Question

[iriskye]

I first mapped to mm10 with bowtie and them did peak calling with macs using genome size 1800000000. Returned 6,632 regions with 3100+ for those high score peaks.

Then I wanted to use the histone modification etc. data from encode but those are in mm9 so I repeat everything using mm9 with all the other parameters the same except the genome size 1865500000 (which I assume it's not much difference) And I chose 'Perform the new peak detection method (futurefdr)' for mm9. However only 3,535 regions returned with high scores at ~1500.

I am confident for the mm10 analysis. But can't understand why mm9 result differs so much.

Thanks to anybody who can help me with that.

[Friederike]

what happens if you run MACS with the same peak detection method (not the "futurefdr") on mm9?

and why are you convinced that the 6,600 peaks for mm10 are fine, but the 3,500 for mm9 aren't?

[GouthamAtla]

Use exact commands while running macs, except the genome size. Instead of looking at the numbers, can you compare the peak coordinates, using tools like LiftOver ? Then find out which are missing and look the missed regions in genome browser and check if you can figure out something.

[Michele Busby]

Are you using a control mapped to the appropriate reference?