question on sequencing strategies and batch effects

Question

[anagonzangel]

Hi all

I have a project in which RNA Sequencing is necessary. Last year, I did a pilot project with 16 samples using an Illumina HiSeq 2500 instrument in Rapid Mode. I did the RNA extractions and quality control myself. Libraries were done at my institution's core facility, with the TruSeq Stranded mRNA protocol because I work with a non-model organism, and my samples were collected in the field. Reads were 100 nt long and SE. Reads per sample ranged from 7 to 22 million reads, and this has worked well for my preliminary analyses.

Now, I am planning to expand the sequencing, but I am concerned I may get strong batch effects if I keep processing my samples “bit by bit”. However, processing all samples (approx. 96) at the same time is very expensive, so I need to consider other alternatives. One possibility that is cost effective is to reduce the read length to SE50. Another possibility is to use a different instrument, like the HiSeq 4000 which yields more reads per lane. Some facilities offer the Kappa Stranded mRNA libraries prep protocol.

I would like to be as consistent as possible and I was wondering if the community has any advice/experience on reducing the number of reads, and/or changing the library prep, and/or changing the instrument in terms of technical variability and batch effects.

Thank you,

Ana

Answer 1

[WouterDeCoster]

Something I did for a project with transcriptomics of 96 samples was to create one pool of all libraries and sequence this pool twice (or more, depending on your machine and needs). This way, batch effects in sequencing are cancelled out because the exact same pool is used every time.

Answer 2

[BIOTECH.DEEPTI911]

Hi folks,

I am transcriptomics data with control and two different samples from liver cell line. The samples were pooled because of financial constrains. The basic pooling strategy was adopted as follows: 3 controls= pooled to make 1, 3 treatment1= pooled to make 1, 3 treatment 2= pooled to make 1, So now I have 3 samples which I sequenced on illumina HiSeq2500 150*2 paired end chemistry with 25 million read depth. I want to know that is this a normal strategy or will I be able to convince my reviewer during paper that I had pooled samples. This sort of work plan made me to lose my confidence. Please help me

[Kristoffer Vitting-Seerup]

Please make your own seperate question instead of writing in another question.