View Single Post
Old 08-03-2010, 12:30 AM   #4
nasobema
Member
 
Location: Germany

Join Date: Jul 2010
Posts: 14
Default

Quote:
Originally Posted by greigite View Post
Not necessarily- expression analysis packages like edgeR and DESeq have library size correction factors built in to the statistics. If one library is severely under-represented though it will reduce the statistical power of your analysis.
Expression analysis is indeed one of the applications that we planned.
If the unequal distribution of barcodes is genome wide, I wouldn't expect severe problems because normalization would be simple. But if there's unequal distribution within the genome (e.g. gene A shows higher coverage than gene B for barcode 1 but lower coverage for the same sample using another barcode) it's more difficult.
A probable solution could be to use two barcodes for each sample and take average values or to limit the number of barcodes used to keep overall coverage high.

I should add that we are working with bacterial cDNA, thus the reference genome size is not very large, which increases the coverage.
nasobema is offline   Reply With Quote