Is this a valid way to transform recombination rate values to 10kb windows?

[javibio-git]

Hello,

I am trying to compare recombination rates for two populations using the same reference genome, but the problem is that pyrho output is essentially different given the difference on the intervals. I think one workaround is to transform the pyrho output to 10[50]kb windows for each population to make them comparable. The main problem is that I am not sure if the way I am doing it is actually valid. For example:

Let's say we have these intervals with their respective rho values:

Interval A: 1-5,000 bp,     rho = 2.0e-8
Interval B: 5,000-8,000 bp, rho = 1.5e-8  
Interval C: 8,000-12,000 bp, rho = 3.0e-8

Now I'd like to calculate rho for the first 10kb window (1-10,000). And the way I should be doing this is first to know the amount of overlap of the intervals above within 1-10kb. This would be A = 5,000; B = 3,000; and C = 2,000. Taking this into account, the calculation should proceed like this:

window_rho = (2.0e-8 × 5,000 + 1.5e-8 × 3,000 + 3.0e-8 × 2,000) / (5,000 + 3,000 + 2,000)
                     = (1.0e-4 + 4.5e-5 + 6.0e-5) / 10,000
                     = 2.05e-4 / 10,000
                     = 2.05e-8 

Essentially, I am calculating the weight average for all the intervals that "contribute" to a specific window. In my head this makes sense but I am not sure if this is correct.

Any insights on this would be much appreciated!

Javi

[jeffspence]

Hi Javi,

I think what you propose makes sense to me, and is essentially what we did in the paper to compute correlations at broader scales (e.g., 1kb or 10kb scales).