2
$\begingroup$

I would like to create an estimate for the expected number of types of gene repeats when drawing from a set of genes where each gene is unique and has equal probability.

i.e.
I have a genome of size N
I have a sample of size m genes
Each gene has a probability 1/N
I define a duplicate as seeing the same gene sampled twice, triplicate thrice, etc
So, from a subset {2,2,3,4,8,9,9,9,9,1,1} taken from some set, you see 2 duplicates, 0 triplicates, and 1 quadruplicate

If I sample with replacement, how many duplicates would I expect to see? How can I expand this to create a distribution of expected pairs, triples, quadruples,..., ntuples

Thank you!

$\endgroup$
1
$\begingroup$

If you have $N$ equally probable genes and sample $m$ times with replacement then the probability of seeing a particular gene exactly $k$ times is (from the binomial distribution): ${m \choose k}\frac{(N-1)^{m-k}}{N^m}$ and so (by linearity of expectation) the expected number of genes appearing exactly $k$ times is $${m \choose k}\frac{(N-1)^{m-k}}{N^{m-1}}.$$

In your example $m=11$. If $N=9$ then the expected numbers for different $k$ are about

 k  Expected number
 0  2.463569
 1  3.387408
 2  2.117130
 3  0.793924
 4  0.198481
 5  0.034734
 6  0.004342
 7  0.000388
 8  0.000024
 9  0.0000010
10  0.000000025
11  0.00000000029
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.