Why using short reads?

[dbrg77]

Hi, everyone. I'm a newbie here.

I have a question that always baffles me that why NGS only uses short sequence tags?

Instead sequence the whole DNA fragments, it only sequences the first 25-50 bp of a DNA fragment.

Is there a particular reason for doing this?

Thank you!

[krobison]

It's what the technology is capable of. Most technologies attain their signal from a clonal population of DNA molecules which are signalling in sync; as this synchronization is lost (dephasing) the noise climbs; eventually noise overwhelms signal. The best review on this is in The challenges of sequencing by synthesis., whose author list features many of the heavy hitters in the field.

PacBio's SMRT technology (and Helicos' before it) looks at single molecules and therefore does not have a dephasing problem. For PacBio, the problem is apparently that the laser used to observe the sequencing eventually destroys the DNA polymerase that is doing the work, probably via a free radical mechanism. I don't know if Helicos ever figured out what kept their reads very short.

Note that short is not quite as short these days; SOLiD is getting around 75bp and Illumina on the order of 150 bp. 454 routinely gets 400+ (perhaps soon out to 800) and PacBio 1.5Kb. Still, all sorts of blue sky proposals are aiming at reads of 10s or perhaps 100s or even 1000s of kilobases.