Hello,

I have a question regarding the normalisation procedure with QSEA. For transformation to beta-value, as well as for normalisation to NRPM, a factor called "sequencing preference" is mentioned. However, I don't fully understand what this sequencing preference is about. It can be calculated with findSeqPref() and added to an existing QSEAset with addSeqPref(), but going through all the scripts and the manual for QSEA, these to functions are not used by default to create the QSEAset. Looking at the final QSEAset, the seqPref is indeed not included in the QSEAset (btw, the code for addSeqPref() does not run smoothly for me, anyway).

Could you explain to me the underlying background of the sequencing preference? Is the sequencing preference considered for beta-Transformation and normalisation to NRPM by default (if not, why?) or is it only considered, if the addSeqPref() function is particularly used/added (if this is the case, what are reasons to particularly choose to add the sequencing preference for normalisation?)?

Thank you!!

Hi Anja,

initially, I added this additional normalization factor to compensate for local sample specific biases. However, in my data I did not see such biases, and therefore the implementation got stuck at an experimental state. The sequencing preference factor probably is not necessary for your dataset, except you did observe any sample specific local biases (other than from CNV) in your data. However, in this case, you would have to think about appropriate normalization methods.

I apologize for any confusion that may have been caused. I will make a note to improve the documentation for these functions in the future, and I appreciate your feedback on this matter.

Best, Matthias