Entering edit mode
7.3 years ago
iibrams07
▴
10
Hi
i am planing to do very soon mRNAseq on different combinations of multiple knockouts (KO) in a set of genes. In total i have 15 different knockouts and i would like to analyse the differential expression of each of them compared to WT (wild type).
I have at my disposal 3 different clones of each of these knockouts as biological replicates. So in total there are 15 x 3 (KO samples) + 3 (WT samples) = 48 samples ( in the case i dont consider technical replicates ).
- Whats the most reasonable way to distribute the samples in the flow cell ? In other words, to be cost effective without compromising the data quality, whats the maximum number of samples to be put in a single lane ?
Is it necessary to include technical replicates ? In this case i would have to deal with about 100 samples !
I am new in the field of NGS and couldnt find an appropriate answer to my concern. Thanks a lot.
(1) This depends a lot on your organism. With humans/primates, we can usually run 8-12 RNAseq libraries per lane, but the actual amount you'll want should depend on your organism (particularly the exome size) and the depth of sequencing you need. (2) Usually we don't use technical replicates (they're essentially just sequencing the same sample to higher coverage), but you do want to make sure your samples are randomized across lanes/flow cells to control for batch effects. How exactly you want them configured will depend on what analyses you plan to do. (3) Your best reference for both of these questions will be what other people working on the same organism/closely related species are doing. If they're using technical replicates, you should probably do so as well; if they're sequencing to 20x coverage, you should too. All that data should be available in the SuppInfo of any papers out there.
Thanks for your answer. In my case the organism is mouse. In the lab i am working in this is the first time we are using NGS. In the case i randomize all the 48 samples and i would like to get 20x coverage and if it is about mRNAseq using paired-end reads, in how many lanes should i distribute the 48 libraries ?
Our lab usually gets 160-220 million reads per lane, so you can work out how that converts into coverage for the mouse genome and then how many samples per lane. You could also look at what other people have published from mice, even if your lab is new to NGS
In your case, whats the average length of the reads ? Thanks.