RNAseq experimental design

[Fitzcarraldo]

Hi everyone,

I am new to RNAseq analysis and am making preparations for the design of an RNAseq experiment to determine the differentially expressed long non-coding RNAs between virus-infected and uninfected cells in culture.

I would greatly welcome comments on what is optimal for the following or indeed what else I should be considering:

1. RNA prep with DNase I (on-column or in solution digestion?)
I am using Direct-zol RNA prep from Zymo Reserach with on-column DNAase I digestion.
2. the library prep
Illumina TruSeq stranded mRNA library prep kit is the plan. Is oligo-dT capture okay?
3. what platform is best?
HiSeq and fast run mode was suggested
4. what number of reads/per biological replicate is needed
Is 3 biological replicates and 10 M reads per biological replicate enough?
5. single read or paired end reads
Single end is okay?
6. read length
>75?

I would be extremely grateful if members could please recommend what kits are best suited to the task as well: Illumina kits? (library preparation kit and ribosomal RNA depletion?) to maximise the chances of success to detect differentially expressed long non-coding RNA transcripts.

Many thanks for your time and expertise and it really is greatly appreciated,

Fitzcarraldo

[hideandSEQ]

Most of these questions are going to be answered based on your budget and the scale of your experiment.

If you want to look for rare transcripts, higher sequencing depth per sample will be more important. If you plan on doing lots of samples with many replicates then you may be less interested in reads per sample.

If your goal is purely to look at lncRNA, oligo-dT capture may not be ideal as many - perhaps even most - lncRNA are not polyadenylated. I've never tried it but ribosomal depletion is a popular alternative.

If you dont care about costs or time and you're using a highseq 1500 or 2500, then high output mode > rapid run mode because you'll get more reads per lane. There are newer "better" Illumina platforms but I dunno what's available for you or if your institution has troubleshooted the new platforms well enough to trust them.

Paired-end sequencing might be more ideal for you because lncRNA often have hella splice variants and you'll want to read through more than one exon to see those.

If possible, first do a trial experiment and analyze the sequence data before preparing all of the libraries to avoid wasting lots of money building bad libraries or sequencing lots of ugly data. You can even sequence this trial run at a higher depth than usual, then see what happens when you down-sample to different read depths to try to optimize sequencing saturation.

[Fitzcarraldo]

Most of these questions are going to be answered based on your budget and the scale of your experiment.

If you want to look for rare transcripts, higher sequencing depth per sample will be more important. If you plan on doing lots of samples with many replicates then you may be less interested in reads per sample.

If your goal is purely to look at lncRNA, oligo-dT capture may not be ideal as many - perhaps even most - lncRNA are not polyadenylated. I've never tried it but ribosomal depletion is a popular alternative.

If you dont care about costs or time and you're using a highseq 1500 or 2500, then high output mode > rapid run mode because you'll get more reads per lane. There are newer "better" Illumina platforms but I dunno what's available for you or if your institution has troubleshooted the new platforms well enough to trust them.

Paired-end sequencing might be more ideal for you because lncRNA often have hella splice variants and you'll want to read through more than one exon to see those.

If possible, first do a trial experiment and analyze the sequence data before preparing all of the libraries to avoid wasting lots of money building bad libraries or sequencing lots of ugly data. You can even sequence this trial run at a higher depth than usual, then see what happens when you down-sample to different read depths to try to optimize sequencing saturation.

Thank you so much for your detailed response! It has been up on my wall here while I cleared other work then I realised I had never actually thanked you!

Can I please ask is it sensible to consider a company to handle the data analysis as this is a big bottleneck for me. Novogene was mentioned as a reputable service provider. Many thanks in advance.