I have RNA-seq gene expression value for small set of genes in 8 tissues from 28 different individuals. But I have only the RPKM values. Now whats the best way to filter out the low expressed genes from there subset?
That depends on the situation, what is the next step? Filtering genes will not be the end of the analysis, do you plan to do differential expression analysis? What is the experimental design?
Next step will be to do a co-expression and differential expression analysis.
You could also consider to plot the distribution of fold changes to get an idea of what is low and what is highly expressed.
I usually use 1 FPKM as cutoff. This gives the convenience that the log of expression values is always positive making downstream analysis less cumbersome. Definitely arbitrary to take 1 FPKM and values should depend on the situation.
you can use some cutoff like: filter all genes with at least 0.1 rpkm in 90% of your samples. You can play with these values depending on your analysis..