[BioC] VanillaICE

[Robert Scharpf]

Hi Francesco,

The LLR is the log likelihood ratio comparing the called state to the 'normal' state.  In the context of identifying regions of homozygosity  (ROH) from the diallelic genotype calls, the normal state is NOT a region of homozygosity.  The likelihood for the two states is computed from the emission probabilities (modeled as Bernoulli), and includes a penalty for jumping to and from the ROH state given by the transition probabilities.   The exact values of the two transition probabilities will depend on the distance between the markers where the jump occurs.  

Thanks for the note -- will clarify in the package vignette.

Rob


On Nov 10, 2011, at 11:54 AM, Valerie Obenchain wrote:

> Hi Francesco,
> 
> I'm cc'ing Robert (package author) in case he didn't see this message.
> 
> Valerie
> 
> 
> 
> On 11/09/2011 08:58 AM, Francesco Mancuso wrote:
>> Dear list,
>> I am using the tool VanillaICE in order to find regions of Loss of
>> Homozigosity comparing Normal versus Tumor cells.
>> 
>> I cannot understand in the results what "state" indicates and how I have
>> to consider the LLR value. The vignette does not give an exhaustive
>> explanation.
>> Could you explain me it or suggest where I can find these info?
>> 
>> Many thanks!
>> Francesco
>> 
>> 
>> 
>