Confirmation sequencing technology

[gendxdoc]

In a few months I will be setting up my 3rd NGS sequencing facility -- this time starting from scratch. I have a HiSeq system on the way, but need to select a technology for performing confirmation sequencing.

Do I drop the money to purchase an AB capillary sequencer (ie. 3500xL or 3730) --- or could I use another small scale NGS platform for confirmation ?

We'll be performing exome and genome sequencing using the HiSeq system -- could the identified variants be confirmed using a MiSeq ? Or would it be preferable to using a different technology, such as the Ion Torrent PGM ?

Or would it still be best to use traditional, capillary-based Sanger sequencing ?
I know Sanger sequencing will be around for quite a while, but I'm wondering if I could do confirmation sequencing more cheaply on a small-scale NGS system.

I welcome your opinions !
Thank you,
Mike

[GW_OK]

I should think a Sanger system would be the way to go. How many things are you wanting to confirm at a time? (Per patient?) NGS might actually be more expensive...

[henry.wood]

I've heard of people using a sequenom, but I don't know much about them. I'd be interested if anyone has any thoughts on them.

[razibus]

If you get a high number of mutations to check, it will be very costly and time-consuming to use Sanger Sequencing but if the number of mutations is reasonable then go with 3730.

As you stated, it is better to use an other technique to validate your mutations. If you get a high number of mutations to check, you can buy a PGM or GS-FLX but it won't help you so much since you will have to make one amplicon for each of your mutation. The solution is to use a robot to perform this task like Raindance or Fluidigm.

There are also other solutions than sequencing, for example take a look on BeadXpress or Sequenom as said Henry.

[pmiguel]

What will you use the mutations for? If you will do genotyping, I guess you could confirm the mutations on the system you were planning to use to do genotyping. Goldengate?

Seems crazy to use a system with a much higher error rate (eg, Ion Torrent) than the HiSeq to do confirmation.

I agree this should be a job for Sanger sequencing. But Applied Biosystems has done basically nothing to advance the state of the art in Sanger sequencing since the advent of second generation technologies. They hold the monopoly on that (shrinking) first generation sector and appear content to let the price per base cost drift upwards.

--
Phillip

[razibus]

Originally Posted by pmiguel

Seems crazy to use a system with a much higher error rate (eg, Ion Torrent) than the HiSeq to do confirmation.

Very interesting. Where did you get the info? Did you read it in a recent paper?

[pmiguel]

Originally Posted by razibus

Very interesting. Where did you get the info? Did you read it in a recent paper?

Well, I read the Ion Torrent forum on this site. Lots of talk about this issue. For example this thread:

http://seqanswers.com/forums/showthread.php?t=13070


Seems like sequencers are falling into the "fast, good, cheap; pick any two" rule. Ion Torrent is fast and cheap, but you pay a price in quality. Well this ignores various stratifications of cost -- like cost to do a single run vs. instrument cost vs. per sample cost. (Both in cash and time.) But seems accurate enough for a rule of thumb...

--
Phillip

[westerman]

On his blog, Keith Robison has a new post about the Ion Torrent and quality vis-a-vis the MiSeq:

http://omicsomics.blogspot.com/2011/...-at-miseq.html

In my opinion, Quality Values are a subjective issue. Everyone wants to tout their machine has having high QVs yet who sets those QVs? The manufacturer does. Which makes the QVs suspect. We really need independent verification. Which is hard to do when the technology keeps changing rapidly.

Keith points out in his blog:

Furthermore, my own analysis shows that Ion is still calling under-calling phred scores. A observed-called plot (below) for the long read dataset for called phred=17 is shown; other called qualities show similar values: Ion is consistently underestimating phred values by as many as 10 points.

Not a very rigorous study but something to keep in mind.

Going back to the original poster's question, it does seem to me that using a different technology than the hiseq-miseq one would provide for better verification. Certainly the Sanger sequencing is known and cheap for sample prep. We still do a lot of it even if, as Phillip says, price per base drifts upward. The non-sequencing solutions are also appealing.