I'm starting to learn meta analysis on R and I've got two questions:
1. If I have two GSE datasets from two different platform, say, GSE1 from GPL1 and GAE2 from GPL2. I'm wondering is it safe to combine their individual p-value to calculate meta-p and meta-weight?
2. I'm wondering can one gene have more than one gene symbol in the NCBI datasets?
This list is generally intended to help people figure out how to use Bioconductor packages, rather than explaining the finer points of statistical analysis, so this is probably off topic. Plus, it's not clear what you mean by 'safe', nor 'combine', so I don't think anybody could answer that question anyway.
This is probably off topic as well, but the answer is a conditional yes, depending on what exactly you mean by 'gene symbol'. There are the HUGO gene symbols, which are intended to be unique, but in my experience are not 100% unique. In addition, there are what we call 'aliases', which are the various different names that people have given genes over the years. These are most certainly not unique, and depending on where you got your symbols from, you may be talking about an actual HUGO symbol or just some random alias that somebody called a gene that one time.
The gene symbol I referred to is the symbol from platform annotation from NCIB (for example: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GPL570, GPL annotion contains gene symbol). I checked HUGO symbols and found that gene symbol from GPL annotation quite match the HUGO requirement. I suppose the gene symbol in GPL annotation is in HUGO format, correct?
The GPL that you link to has the Affy annotations from 2011. While the gene symbols they list there may be HUGO gene symbols, if you plan to compare to a data set that was annotated in 1995 or 2015, there may well have been changes in HUGO itself, where some gene symbols got dropped, or changed or whatever.
If you are talking about Affy arrays, you can always either use the Bioconductor annotation packages, which contain the most recent mappings, or go get the annotation csv files from Affy and use them. But relying on a static upload from whenever might not be the ideal strategy.
Thank you very much for your help James!
For question 2:
The gene symbol I referred to is the symbol from platform annotation from NCIB (for example: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GPL570, GPL annotion contains gene symbol). I checked HUGO symbols and found that gene symbol from GPL annotation quite match the HUGO requirement. I suppose the gene symbol in GPL annotation is in HUGO format, correct?
Thank you
The GPL that you link to has the Affy annotations from 2011. While the gene symbols they list there may be HUGO gene symbols, if you plan to compare to a data set that was annotated in 1995 or 2015, there may well have been changes in HUGO itself, where some gene symbols got dropped, or changed or whatever.
If you are talking about Affy arrays, you can always either use the Bioconductor annotation packages, which contain the most recent mappings, or go get the annotation csv files from Affy and use them. But relying on a static upload from whenever might not be the ideal strategy.