Will you stop using 454?

[james hadfield]

I wonder how long Roche 454 will still be around if Ion gets to 400bp. Also if Illumina's chemistry performs as well as it looks like it will then 200-300bp is not unreasonable, coupled to PE reads we could see Illumina at 400bp mini-contigs.

With two short-read platforms offering close to Roche 454 read-lengths but with higher numbers of reads, lower costs, and in Illumina's case the simplicity of sample prep, when will we stop using 454?

[nickloman]

We'll probably keep using it for amplicons (particularly 16S) until a cheaper platform gives reasonably accurate long reads above 400bp. Ion Torrent are promising this sometime in 2012. Illumina may get there around the same time e.g. a 2 x 200bp paired read is not too difficult to imagine. This would take a long time on the HiSeq but perhaps not on the MiSeq.

[maubp]

But Roche won't stand still - I recall hearing something about 1kb read 454 upgrade being tested... does anyone know more?

[lletourn]

We've been using, as beta, their long reads and they work fairly well. I can't give numbers but we had reads that hit the 1kb mark.

Even at 700b the median was over Q20.

Roche is not standing still for sure. When Ion torrent hits the 400b, I'm gussing 454 will already be at the 1k, 1.2k mark.

[avilella]

I wonder how long Roche 454 will still be around if Ion gets to 400bp. Also if Illumina's chemistry performs as well as it looks like it will then 200-300bp is not unreasonable, coupled to PE reads we could see Illumina at 400bp mini-contigs.

With two short-read platforms offering close to Roche 454 read-lengths but with higher numbers of reads, lower costs, and in Illumina's case the simplicity of sample prep, when will we stop using 454?

454 plans to launch a new sequencing chemistry for the GS FLX that will reach read lengths to 750 bases and beyond. Supposedly by the end of June...

[james hadfield]

750-1000bp reads are very useful but with only 1-1.5M reads is it better sticking to 200-400bp and getting maybe 3-4M of them on MiSeq?

I do not run any long read projects, we did a couple of Metagenomics experiments where long reads were more helpful. However a groupjust published a comaprison of 454 to Ilumina for this and the increased number of reads seemed a pretty strong reason to use Ilumina. This did not take account of possible 200-400bp mini-contigs.

What applications do you need 1000bp or more for that can't be broken down into short uniquely barcoded amplicons tiled across regions?

None of these companies stand still.

That's what makes our jobs so much fun!

[lletourn]

16s metagenomics...instead of having individual regions, you'll get almost all regions in each reads. Great for measuring diversity, richness, etc.

Also longer reads really make a difference when assembling. Mate pairs make a bigger difference, but when combining both...

[kmcarr]

What applications do you need 1000bp or more for that can't be broken down into short uniquely barcoded amplicons tiled across regions?

Plant transcriptomics. With so many recent segmental or whole genome duplications in plants assembling paralogous transcripts from shorter reads can be maddening.

[RCJK]

I've been told the Titanium XL+ upgrade should be launched by the end of May, but don't have any more details than that other than that new computing is required to process the data.
It's taken Roche way to long to launch this and they are too quiet regarding any other plans they may have. I believe they have licensed the same semiconductor tech as Ion Torrent so I'm waiting for an announcement one day of a similar platform....

[Michael.James.Clark]

Plant transcriptomics. With so many recent segmental or whole genome duplications in plants assembling paralogous transcripts from shorter reads can be maddening.

How much will 1kb reads help with that versus 400bp reads?

There are probably better solutions than longer reads for such situations. For example, unique tagging every transcript in individual droplets during library generation.

[petang]

How about PacBio machine that claimed a maximun read length of 3kb with an average of 1kb? The only problem is they can only generate ~80000 reads on a 30 min run.

[niceday]

it would also be useful if roche automated their emulsion pcr-breaking and enrichment steps. SOLiD have the ez system, ion torrent will have the one touch. Doing any of these processes manually is a right pain. I'm suprised roche haven't done something about this.

[krobison]

The other player in this space will be PacBio; their reads are apparently pushing 5Kb. If you are primarily using long reads to scaffold short ones, then the ~80% single pass accuracy may not be an issue.

Of course, getting an instrument will set you back a lot more than any of these others . But, there seem to be enough in the field that it is likely one could get access for a project without buying one.

[GW_OK]

It's taken Roche way to long to launch this and they are too quiet regarding any other plans they may have. I believe they have licensed the same semiconductor tech as Ion Torrent so I'm waiting for an announcement one day of a similar platform....

I absolutely agree. I remember them first talking about 1k reads way back in 2007-8 or whenever Titanium was first coming out. It was always going to drop in the next 6 months and was then always pushed back.
And we were always, always asking about when they were going to automate emPCR. I wonder if they can license the OneTouch from iotorrent, lol.
The 454 is quickly becoming the has-been of the field.

[clivey]

What if

Your reads were over 100kb, very accurate and you had a mountain of them ?