Well, If you are planning to outsource sequencing, first you should consider the budget you are planning for sequencing…..then you should consider platform.Because there are lots of platforms and again there are systems under each platform… Currently I guess Illumina Hiseq 2000 is best, because it gives a very deep sequencing. Most of the downstream analysis software also support well for illimina.

I am not too familiar with 454 sequencing but I would assume that you would do absolutely fine with Illumina sequencing. I think the number of reads that you get out from one sequencing run is higher than for 454 (which would allow you to pool many samples in a sequencing lane at a pretty low cost). Since you want to profile miRNAs, which are very short, you would not benefit from the much longer reads of 454 sequencing anyway. I would most probably go with Illumina HiSeq2000.

For miRNA profiling, Illumina or SOLiD technologies are clearly better because they produce reacher datasets of short reads. Both allow multiplexing. Both have their pitfalls. Which one suits you better - depends on your purpose and resources. E.g. how deep profiling do you intend? Would you focus on low-expressed species? Will you search for novel species? etc. I assume sequencing will be purchased as a service, but what about bioinformatics? If you intend to use it for diagnostics rather than discovery, perhaps array would be a better cost-effective option?

Lisandro, the only advantage 454 has is read length (and even that advantage is starting to evaporate). Since your application is miRNA, there's absolutely no reason to use 454 (unless it's the only technology you have access to). If you have a lot of samples, Illumina's HiSeq would likely be the most cost effective solution, but it will require multiplexing many samples together in a single lane. (SOLiD can also produce good results cheaply, but Illumina has clearly won the battle of the high throughput systems). If you don't have that many samples, one of the desktop solutions could work, either the MiSeq (also from Illumina) or the PGM or Proton from Ion Torrent.

Good luck and let us know what you decide!

Lisandro,

I agree with Shawn and Martin. You can multiplex 12 samples on an Illumina Hi Seq 2000. We recently did 8 samples on a single lane of a flow cell and ended up with great depth and a lot of data.

The costs are coming down also. I recommend you shop around to get the best quote!

Good luck

I also agree - Illumina is the best - generates very high quality reads. I am currently busy with our library prep for miRNAs to sequence on the HiSeq 2000. If you don't have a lot of samples - perhaps the MiSeq might also be an option.

Due to the low level of miRNA in body fluids I find library prep challenging. What we tend to do it MiSeq it first and if we are happy with the quality we run HiSeq. Would you be able to tell me what is the volume of plasma you use? Thank you.

Thank you everybody for your reply. My goal is to profile circulating miRNAs in plasma as biomarkers of kidney injury. First, I want to do deep sequencing looking for miRNA's in a unbiased way in 32 patients (4 groups of study) . Then, I'll run real-time PCR to detect the upregulated mirNA's found by deep sequencing in a greater population (200 patients). I hope you understand to me. So, the sequencing technique is very important, and Illumina appears to be very useful for my goal. Thank you very much for to guide me. Is my first experience with deep sequencing and your advices are really important. Dear Anna Hoy, I've been reading that 400-500 uL of plasma is enough to yield good quantities of RNA of low size. I'm begining my study (taking the blood samples, 20 ml for patients).

Thank you again!

Hi Lisandro , I just wanted to warn you about your typical approach with a small dyscovery set and a large validation one. This is typically designed by clinicians in order to save research money. There is nothing worst than a small dyscovery set to destroy your experiment. While people is generally concerned about false positives and anybody understand that they can be cleared out in a subsequent validation experiment, little concern is shown about false negatves. They represent true positive that will be excluded from further analysis in your (small sized) dyscovery experiment and will not be rescued by the validation one, and are lost forever. It is really more meaningfull to make a large dyscovery experiment, after that you will need only a few independent sampkles to validate the findings. It is more expensive but the "wedding with dry figs" approach has never workout that well.