How to define Basal B subtype from TCGA breast cancer patient samples ?

Question

[ZheFrench]

TCGA provide a molecular classification (Luminal A , Luminal B, Basal, Normal like, HER2 enriched) Basal subtype can be divided in basal A and basal B. (Kao et al., 2009; Neve et al., 2006) It has been done using expression from MicroArray via Significance Analysis of Microarrays (SAM) or PAM using the nearest shrunken centroid methodology. I retrieved from the paper a gene list used to classify so I have some ideas to begin to work but I was wondering if there ware a more standard approach to do that using TPM gene expression values.

[Kevin Blighe]

I would simply recommend clustering the samples using the gene signature that you have obtained. You could try the TPM values and also the scaled TPM values.

[ZheFrench]

Yeah that was my first idea. By the way how do you scaled TPM values ?

[Kevin Blighe]

If you perform clustering using heatmap or heatmap.2 (in R), then just set scale=TRUE. Otherwise, you can scale your data to Z-scores with t(scale(t(MyData))), assuming that MyData has samples as columns and transcripts as rows.

Usually, people cluster, i.e., perform hierarchical clustering, the non-scaled data and then present the scaled data in the heatmap.