1
$\begingroup$

I'm a computational biologist trying to interpret this equation correctly.

$$CI(x)= 1- \frac{2}{(N(N-1))} \sum_i^N\sum_j^N [ S(x_i,x_j) / \sqrt{S(x_i,x_i)S(x_j,x_j)}\ ] $$

i=1 ; j=i+1

I'm confused about the part before the summation

This is how I under stand this:

This is a nested summation, so if N=3 then you sum the partitions like[(1+2)+(1+3)]+[(2+3)]

Then do you multiply by $ 1- \frac{2}{(N(N-1))}$ ? or are you multiplying the sum of i (and j?) for each summation cycle, such as multiplying for 1 and 2 instead of multiplying 1+2?

Thanks a lot, it's been a while since I worked with stuff like this. This is a conservation index metric for multiple sequence alignments using a BLOSUM62 substitution matrix (The S(x,x) part).

$\endgroup$
  • 1
    $\begingroup$ First, calculate the double sum. Then you multiply by $2/(N(N-1))$, which is a fixed number, since $N$ is fixed. Finally subtract all that from $1$. $\endgroup$ – angryavian Feb 27 '14 at 4:07
  • $\begingroup$ THANK YOU SO MUCH! I feel so dumb now. $\endgroup$ – QVINTVS FABIVS MAXIMVS Feb 27 '14 at 4:11
1
$\begingroup$

Just focus first on the summation; it write $$\sum _{i=1}^N \sum _{j=i+1}^N \frac{S(x(i),x(j))}{\sqrt{S(x(i),x(i)) S(x(j),x(j))}}$$ For clarity, I shall expand this sum for $N=4$ in order you see how the terms appear $$\frac{S(x(1),x(2))}{\sqrt{S(x(1),x(1)) S(x(2),x(2))}}+\frac{S(x(1),x(3))}{\sqrt{S(x(1),x(1)) S(x(3),x(3))}}+\frac{S(x(2),x(3))}{\sqrt{S(x(2),x(2)) S(x(3),x(3))}}+\frac{S(x(1),x(4))}{\sqrt{S(x(1),x(1)) S(x(4),x(4))}}+\frac{S(x(2),x(4))}{\sqrt{S(x(2),x(2)) S(x(4),x(4))}}+\frac{S(x(3),x(4))}{\sqrt{S(x(3),x(3)) S(x(4),x(4))}}$$ Once this summation is done, just multiply the sum by $ \frac{2}{(N(N-1))}$ and substract the result from $1$.

$\endgroup$

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.