Hello,
my colleague sequenced (RNA-Seq) some samples a month ago and the quality of some of them was not so good. She sequenced some new samples and also some of the bad quality samples again to analyze them again. She is asking me if, in order to have more reads and better results, I can re-use the bad quality old reads (.fastq files) and merge them with the new ones (only with the samples that have been re sequenced). Is this possible? Is it recommended or bad practice? It does not sound very good to me, but I don't have experience with this.
If it is a good idea, should I do it with command 'cat' or how? thank you.
•
written
by
Pin.Bioinf
In my opinion this does not seem like a good idea. If your quality is bad, don't use it. If you really would like to use technical replicates in your RNA-seq design (and ignore the fact that you have bad quality), you can use the duplicateCorrelation function in limma to include the technical replicate information, so no don't use cat command here either.
Okay, thank you. Maybe so my colleague agrees that it is not a good idea I could check if the replicates are consistent plotting a PCA (and probably this will show they are not similar) so then I can convince her it is not a good idea?
Quantify the genes from both technical replicates and check the correlation between two samples. PCA might not give a good idea between technical replicates. Its good for biological replicates, as it takes only top 500 or 1000 genes. "Bad quality samples" either mean the initial RNA is of poor quality, or low starting material etc. In this case its better not to use the replicate.
If bad quality means, drop in base qualities or low sequencing depth etc, then you can use them as the gene expression quantification are not going to change much (After quality trimming).
Log in to answer this question.
You don't want "more" reads, you want representative numbers of reads.
For genome assembly, sure, it would likely be fine. Not for an application where read numbers count though.