What is control_lambda?

[Mariam]

Hi,

I was wondering if someone could explain to me exactly what I'm looking at when I look at the outputs from the peak calling. In my outputs I get sample1_treat_pileup.bdg and sample1_control_lambda.bdg and I was wondering if the numbers in the pileup I get reflect the number of reads aligned to each chromosome region in the ChIP? Is the control lambda a proxy for number of reads aligned in the INPUT? I'm trying to compare two groups of samples and see if they show differences in amount of signal in a certain region. The signal is weak in general and doesn't show up as a significant peak but I was hoping I could use the pileups and a large group of individuals to run some stats and show a consistent difference in enrichment.

Thanks,

Mariam

[Ian Donaldson]

The lambda background model is a "black box" to me, but the best description of it that I have found is on this forum.  Hope it helps.

Ian

https://groups.google.com/d/msg/macs-announcement/JkufzGpUNRk/kUx0z2M2b_cJ

For each peak region, MACS calculates the a local lambda for poisson 
distribution based on the control tags within the 1kb, 5kb and 
10kb(1/5/10k are parameters that you can modify) nearby regions to 
consider the local fluctuations and biases. The local lambda is the 
maximum of the averages of tags for 1/5/10 kb regions and a whole [constant]
genome background. Then this local lambda is used to calculate the p-
value of poisson distribution. If there is no control data, the ChIP 
data will be used instead, where the 1kb region is not considered. The 
fold-enrichment is also calculated using local lambda.

--
You received this message because you are subscribed to the Google Groups "MACS announcement" group.
To unsubscribe from this group and stop receiving emails from it, send an email to macs-announcem...@googlegroups.com.
To post to this group, send email to macs-ann...@googlegroups.com.
Visit this group at http://groups.google.com/group/macs-announcement.
For more options, visit https://groups.google.com/d/optout.