How do you pronounce ‘Gemma’?
The ‘G’ is soft, as in ‘general’.
How do you map probes to genes?
Essentially as described by Barnes et al, 2005. Gene annotations are obtained from NCBI and UCSC.
Isn’t expression data very noisy?
Yes, sometimes. This is a motivation for performing meta-analyses: to look for results that are in some sense consistent across laboratories.
Isn’t data quality a problem?
Yes; see the question about noisy data. We have been working hard to ensure that data sets we use for analysis are of high quality, or to “clean up” those that have problems. One problem we have observed is the presence of outlier samples, which are flagged and removed. Batch correction is implemented where possible and we analyze data from raw sources (CEL files or FASTQ files) where possible.