Is the consistency of the NGS library size important?

Question

[mrs.hope]

Hello everyone! I have a general question about the importance of keeping the library fragment size consistent between samples. All my libraries have fragments ranging from 200 to 1000 bp according to the Bioanalyzer, which is fine for the Illumina paired-end sequencing. However, some of the libraries have average size of let's say 300 bp, others - 350-400 bp. I was wondering if this could be a problem by introducing some kind of bias? As far as I can see as long as the alignment rate is the same between the different samples, there shouldn't be any other hidden source of bias. In my particular example, I am doing spiked-in RNA-seq and ChIP-seq experiments and I need to prepare libraries from sonicated genomic DNA to find out the ratio of drosophila to mouse reads in the sample to determine the consistency of the spike-in. For some reason, the sonication efficiency varies slightly from sample to sample, which results in the libraries that have slightly different fragment size distribution. I would really appreciate if you could share with me your opinion/experience/piece of advice on this matter!

PS Crossposted on seqanswers to get more feedback

Answer 1

[Carlo Yague]

I don't have a full answer to this question but since you ask for any bit of advice, here comes my two cents :

• Variation in the sonication efficiency from sample to sample, even in the same batch, is not unusual. At least it happened to several people i know. They choose to throw away the samples that differed too much from the others before doing the library preparation, but small variations (like you have) were considered acceptable.

• A way to take fragment size into account for ChIP-seq experiments is to shift reads on each strand by half the fragment length.