News for hackers. Stuff that matters.
Re:Computational feat speeds finding of genes to millisecond... by carbocation on Thursday April 08, 2010 @01:04AM (#1249185)
It seems to me that the number of possible connections between genes should be (16000 genes * 16000 possible gene-gene interactions per gene) = 2.6 * 10^8. Why do you say 3 * 10^5?

Also, while a 14% FDR seems pretty good, it's just from one sample. I would like to have seen them repeat this procedure on (random) different pathways so we could see how variable the FDR might be.

Finally, I don't know how much this is truly enriched for B-cell genes as compared to a randomly seeded query. I'd like to see them compare their results to what they'd get if they had seeded their query with 2 random genes. Perhaps you could build a story about the association between a large fraction of all genes and B-cell development...

In short, this seems inadequately controlled.

There are no comments on this comment yet.