To uncover specific traits of the exome expression in hematopoietic cells, we compared whole blood samples and CD34+ purified hematopoietic stem-progenitor cells (HSPC) to a compendium of 33 samples from 11 solid tissues.
Based on a microarray compendium of 205 samples, we compared the gene expression profile of mature MII oocytes and human ES cells (hESC) to that of somatic tissues.
Therefore, we generated a transcriptome compendium of 205 samples by collecting public microarray data and compared the gene expression profile of oocytes and hESC to that of somatic tissues.
Predictive GBA analysis for 17/18 tested lincRNAs was conducted as follows: Pearson correlation values of FPKM expression profiles were calculated for each lincRNA to all protein coding genes across a compendium of RNA-Seq samples (combination of in-house samples and samples from [ Merkin et al., 2012]).
We also obtained a compendium of 336 expression samples in human B cells from [ 6], which includes samples from human blood, cancers, and cell lines based on the HG-U95Av2 Affymetrix arrays.
We will focus mainly on two datasets: an expression compendium of cancer tissue samples [ 12] and a study on MCF7 cells where secondary target activation has been blocked by addition of cycloheximide, an inhibitor of protein synthesis [ 20].
Tiles are filled based on the Z-score of the Pearson correlation values between a given lincRNA and the genes within the gene set across a compendium of RNA-Seq samples.
Our results indicate that new breast cancer datasets should not be analyzed in isolation, but should be analyzed in the context of the available compendium of breast cancer samples The resampling strategy has been proposed for the derivation of robust signatures [ 5, 13].
These can be useful since a compendium increases the number of samples, while gene sets reduce the size of the feature space.
For each context c, it counts (i) K, the total number of samples in the compendium that exhibit the pattern, (ii) n C, the total number of samples in context c, and (iii) k C, the number of samples in context c that exhibit the pattern.
