Negative values occasionally show up when adjusting for batch or surrogate values. These are expression values that were likely already very low. If they bother you, feel free to ignore them or set them equal to zero. On Jun 5, 2013, at 3:13 AM, Angela Re wrote: > > > Dear Dr. Parker and Dr. Evan, > > I would like to use the 'sva' package to correct the intensity values in a > multi-platform microarray analysis for batch effects. > > I applied sva() to get the the surrogate variables followed by fsva() to > get the amended version of the initial intensity values. I am not > interested in the setting training/test samples. Thus in my case 'dbdat' > and 'newdat' coincide. > > I noticed that the correct values can take negative values. Do you think > it is plausible? Which are the implications in the following differential > expression setting (which I am interested in)? > > I found the presence of negative correctd values also by applying ComBat. > > Thank you for your kind attention. > Best regards > > Angela Re, PhD > > > Centre for Integrative Biology (CIBIO) > via Sommarive 14 > 38123 Povo (TN) > www.unitn.it/en/cibio > > Mobile phone: (+39) 328 4180132 > Skype ID: ange.re > E-mail: re at science.unitn.it > Linkedin: http://www.linkedin.com/pub/angela-re/b/b58/285