[Funcoup] new PPI metric

[Erik Sonnhammer]

Hi, I tried to fix this in the paper, but it turns out harder than I 
thought.  Could you pls rewrite that para Andrey?

BTW Dimitri says he can fix the formatting issues tomorrow.

/Erik

On 10/11/2011 03:42 PM, Andrey Alexeyenko wrote:
> Yes, the formula is indeed wrong. But not the algorithm, as usually!..
>
> This whole term ln(|PP(A,B,..)|) is assigned value 1 if there were fewer
> preys then 3. If 3, then ln() is already >1 (because there are 3
> prey-prey pairs between A, B, and C).
>
> We should correct it in the manuscript.
>
> Thanks for the observation!
> Andrey
>
> On 2011-10-11 14:30, Thomas Schmitt wrote:
>> Hi Andery,
>>
>> you wrote in the FC2 paper that the penalizing term for prey-prey
>> pairs is:
>>
>> ln(|PP(A,B,..)| +1)
>>
>> Where |PP(A,B,..)| is the number of prey-prey relationships in the
>> assay if A and B are preys and 1 otherwise.
>> Have you really used the natural logarithm here? That would mean that
>> none prey-prey pairs would be penalized with ln(2) = 0.693 ?
>>
>> /Thomas
>>