And what is the point in suggestion that : if you want to remove duplicates, try to remove from paired end reads instead of single end reads? what is the difference between them in this case?
Hi all, I have a primary question about preparing RNA samples. We have cancer tissue. We extract the RNA from the cell, then make cDNA from the RNA. We fragment the cDNA and size select. Then we send them for RNA-seq?
If is it like this, when a gene is expressed more, we will have more RNA strands associated to it in our sample. So we would have more reads for that RNA strand(gene) and so on....
So it is possible that we will have more duplicates for a highly expressed gene. If during the analysis we omit those duplicates we will loose information which shows we have had more expression related to that gene?
Thanks a lot in advance Narges
3 answers
I edited your post for clarity. Yes. If you have a high coverage of the transcriptome, chances are you will get a lot of duplicate reads. If you remove these duplicate reads, you will get artificially low expression values.
Hi Narges,
as Dk pointed before, you will have more reads from highly expressed genes, the filtering step can be used to speed the mapping but you need to correct the distribution of reads after that, in many cases it isn't necessary, simply map all your reads with BWA or Bowtie. BTW in paired-end reads you need to remove the pair, no just one read or you will have unpaired (and complex) reads.
thank you so much
Log in to answer this question.
This question has been addressed elsewhere in the forum. See this question and follow its links to other questions in biostars and seqanswers. http://www.biostars.org/post/show/17879/does-illumina-paired-end-reads-from-rna-seq-need-to-be-groomed/#17884