Identifying differentially mutated subnetworks

[jhrcook]

Hello,
Thank you for developing this tool and making it open source. I have a question about how Hefree et al. actually identified the subnets mutated in each cluster. In the online methods, it says that the network-smoothed mutational profiles were analyzed with the non-parametric option of SAM. Were the "network-smoothed mutation profiles" produced on a per sample basis (ie. one per sample, starting with a 1 at each mutated node and RWR-smoothing from there), or were the mutational frequencies of the genes in each sample smoothed? In other words, was a network-smoothed mutational profile made for each sample separately, or was one network-smoothed mutational profile made for each cluster (more like HotNet)?

Thank you for any help you can offer,
Josh Cook

[justinkhuang]

Hi Josh,
If I recall correctly, a network-smoothed mutational profile is generated for each sample individually and then differential values were calculated by measuring the differential values between each cluster and all other samples not in a cluster.