Technical Report BCF-SIB 2007-1

Technical report [PDF 4.4Mb]
Supplementary methods [PDF 77kb]
Supplementary results [PDF 5.8Mb]

The main content of this technical report has been published in Wirapati et al. (2008) Breast Cancer Research 10:R65, along with some modifications to make it more clinically oriented. However, it does contain some additional analyses (although not necessarily relevant to the main point of the published paper). Please cite the published paper if the referred material is already included there. Otherwise, you may refer to this techrep as

Wirapati P et al. (2007) "Integrative analysis of gene-expression profiles in breast cancer: toward a unified understanding of breast cancer sub-typing and prognosis signatures". Technical Report BCF-SIB 2007-1, Swiss Institute of Bioinformatics. http://www.isrec.isb-sib.ch/~pwirapat/bcfsib2007-1

Table of module scores and subtypes

The table of per-patient module score and subtype classifications can be found here. (This is a plain tab-delimited text that can be opened easily from a spreadsheet program.)

Descriptions of the columns:

dataset             dataset identifier
id                  patient id (as in the original public data)
estrogen            estogen-related module score
17q                 17q-related module score
proliferation       proliferation-related module score
p(ER-/HER2-)        posterior probability of being in ER-/HER2- cluster
p(HER2+)            posterior probability of being in HER2+ cluster
p(ER+/HER2-)        posterior probability of being in ER+/HER2- cluster
subtype             1: ER-/HER2-, 2: HER2+, 3: ER+/HER2-, 0: not carcinoma

Notes:

There are (rare) borderline cases where the class probabilities are very similar among competing clusters. For simplicity, the publication uses subtype (last column) according to maximum posterior probability, disregarding the probability values.
The subtype "not carcinoma" is based on information provided in the original annotation. These samples are not included when fitting mixture models.
The dataset EMC2 is not included because the lack of estrogen- and 17q-related genes in the custom 76-gene platform.

Curated Data

The curated data in R .rda files can be found here.

Last modified: 2013/06/19 14:43