Q4.e

[Jose Juan Tapia]

In question Q4.e they ask us to calculate a way to score different subgraphs using a binomial distribution. While the pvalue that we would obtain from such evaluation can be used for comparing between subgraphs as indicated in the homework, it'd also be natural to use it

as a test of significance for the different timepoints in a cluster (e.g. when we calculate individual pvalues for each time point in a cluster, should a pvalue be small enough it means it's not signiciant; for example it could mean that at that particular timepoint it just happened that a lot of genes were being activated, and as such those edges aren't really a good indicator for correlation), so we're not sure if this is also expected from our implementation.

Should we take this into account for our programs? In such case, what would be the required significance cutoff point.

Yours,

--
José Juan Tapia Valenzuela
Carnegie Mellon -- University of Pittsburgh
Ph.D. Program in Computational Biology