We analyzed gene-expression data from six microarray datasets containing up to 12 classes of differentiated thyroid pathologies and normal thyroid tissue.
In our study, we analyzed six publicly available microarray datasets containing 347 thyroid tissue samples and defining as many as 12 classes of thyroid tumors.
A recent study involving 3 independent breast cancer microarray datasets containing a total of 699 patients revealed that CEP55 and FOXM1 are amongst the signature prognostic markers which predicts breast cancer outcome [41].
Nine microarray datasets containing normal tissue samples from previously published studies were collected from the Gene Expression Omnibus (GEO).
All microarray datasets containing values higher than 25 were log transformed, under the assumption this data was non-log-transformed data.
Due to the relatively short observation period in our patient cohort, we used large published microarray datasets containing gene expression data linked to clinical and prognostic information in NSCLC patients.
Additionally, two microarray datasets containing cases and controls in a study of breast cancer (GSE27562) [ 44] and another study of genes with increased expression in AML (GSE9476) [ 45] were used to check the predictive performance of features consisting of circuit activation values estimated using the proposed method.
After excluding probes whose expression values were missing in GSE28521, we identified probe IDs that were mapped to the same Entrez GeneIDs in each set of microarray data and collapsed their expression values by averaging them to make each microarray dataset contain non-redundant sets of genes.
The complete microarray dataset containing 34 samples from the inner and outer cortex, inner and outer medulla, papillary tips, and renal pelvis and from glomeruli is available at https://www.med.stanford.edu/jhiggins/Normal_Kidney/index.shtml.shtml
