Accurate Prediction of cancer outcome helps choosing the best therapy for patients. This site provides access to over 25 gene expression data sets for 13 types of cancer. Try single genes or combinations, assess their predictive value and see whether you can beat the authors of data sets.
If you use the datasets for validation of your methods please cite: Roy J., Isik Z., Winter C. and Schroeder M., Network information improves cancer outcome prediction. Brief Bioinform. 2014 Jul;15(4):612-25. doi: 10.1093/bib/bbs083
These datasets were obtained on the 1th of April running the following query in PubMed.
(cancer[tiab] OR neoplas*[tiab] OR tumor[tiab]) AND
gene expression[tiab] AND
(marker[tiab] OR biomarker[tiab] OR signature[tiab]) AND
(predict*[tiab] OR diagnos*[tiab]) AND
(survival[tiab] OR outcome[tiab] OR progression [tiab] OR response[tiab] OR metastasis[tiab] OR behavior[tiab]) NOT (networks[tiab] OR pathway[tiab] OR review[pt] OR Tissue Array Analysis[MeSH Terms] OR Protein Array Analysis[MeSH Terms]).
The resulting publications were filtered for impact factor greater than 5 (by Thomson Reuters (ISI) Web of Knowledge, version 2009) and if the data was obtained using either Affymetrix HGU133plus2 or HGU133A chips. If raw microarray data was available, we used the RMA implementation as provided in the affy R package, which automatically performs background correction and summarization and then transforms the expression values into log2 scale.
Single-gene marker approved by FDA
|Name/Symbol||Cancer||Link to NCBI|
|Alpha-fetoprotein||liver cancer, germ cell tumours, ovarian cancer|
|Bladder tumor antigen||bladder cancer|
|CA 125||epithelial ovarian cancer|
|Carcinoembryonic antigen (CEA)||colorectal cancer, lung cancer, breast cancer|
|Estrogen receptor (ER)||breast cancer|
|Nuclear Mitotic Apparatus protein (NUMA1)||bladder cancer|
Best single gene marker per data set
|Set||Cancer||Name/Symbol||Link to NCBI||Accuracy|
Signatures published by data set authors