Roche GS Junior mini 454 coming 2010

[DSB]

Roche annouced 19th November 2009 at the Association of Molecular Pathology Annual Meeting in Kissimmee, Florida that they will ship a smaller desktop verison of the 454 in 2010.

Key parameters: (according to their statment

Throughput 100,000 reads per 10 hour run > 35 million filtered bases
Read length: Modal 500 bases, average 400 bases
Accuracy: Q20 (99%) at 400 bases
Comes with desktop computer and software to do some standard processing
Cost will be approx $100K to $125K
Basic sequencing Chemistry - the same as 454.

see www.gsjunior.com

[Zigster]

apparently they figure everyone has forgotten about the IBM PCjr

[nickloman]

I wrote a little blog entry about this:

http://pathogenomics.bham.ac.uk/blog/2009/11/454-junior-announced-a-new-strategy-for-roche/

My feeling is that it might get take-up for a few specific areas; the small bacteriology or virology laboratory interested in de novo sequencing of bacteria or deep sequencing viruses. Perhaps some people doing EST discovery. And maybe clinical diagnostics labs looking to screen amplicon libraries.

[BClan13]

Some questions for the community:

What performance improvements would allow Roche to sell these to a broader number of customers?

Could this kind of platform be, one day, as ubiquitous as ABI 3700s and real-time PCR machines are today?

What emerging platforms could give Roche a run for their money in individual molecular biology labs?

[pmiguel]

Some questions for the community:

What performance improvements would allow Roche to sell these to a broader number of customers?

Could this kind of platform be, one day, as ubiquitous as ABI 3700s and real-time PCR machines are today?

What emerging platforms could give Roche a run for their money in individual molecular biology labs?

The number of ABI 3700s remaining in labs would be very low[1]. I think you mean 3730/3730XL and 3130. My pedantic qualification aside, this is an interesting question.

Ignoring the hyper-Moores law aspect of the sequencer market, what Roche would need to do would be lower their per base cost compared with the Solexa. Longer reads are good, but at a 10-fold higher price per base, the market for the instrument is limited.

But, in fact, it would be folly to ignore the fact that even among the second gen instruments the price per raw base is halving, what?, every 3-6 months? So, who knows. Odds are that 5 years from now all 454 will be door stops...

--
Phillip

[1] This is off-topic, but:

AB struggled during the first couple of years with these instruments. I think they were initially expensive to maintain. Also a lot of slab gel users (377) really hated them. Not sure why. The first time I did a run on a 3700, I regretted ever buying a 377.

Anyway, AB replaced the 3700 with 3730XLs and offered attractive trade-in deals on the 3700. They also stopped development of POP37 on 3700, relegating them to shorter reads than the 3730s.

Overall the 3730s are a superior platform in many ways, but I can't help feeling like the 3700 got short shrift. Remember that at the time of its release, it was Molecular Dynamic's market to lose. Their Megabase had been out in the field for a full year before their first serious competitor, the 3700, was released by Applied Biosystems. Dispite this major disadvantage, the mighty 3700 crushed the Megabase, at least as far as market-share in the US was concerned. Then, Applied Biosystem's foe defeated, how do they reward the warrior class they had created? The replaced them. 3700s are more commonly residence of landfills than laboratories these days.

[nickloman]

But, in fact, it would be folly to ignore the fact that even among the second gen instruments the price per raw base is halving, what?, every 3-6 months? So, who knows. Odds are that 5 years from now all 454 will be door stops...

At this rate, 5 years from now, all the sequencing machines will be door stops!

[avilella]

I would like to ask to the list what are the prospects of the GS Junior for transcriptomics:
probably not in the sense of digital gene expression analysis but more of the kind of research one sees in BMC Genomics, for cDNA discovery in species with no reference genome, which is specially important in plant genomics.

What are the current tools to filter/reduce the mRNA set to interesting medical/agricultural gene families? Considering the GS Junior will probably achieve ~100,000 reads per run, sample preparation will be essential to get the most of each run.

Anyone?

[Ike]

At this rate, 5 years from now, all the sequencing machines will be door stops!

This is an interesting, but I have to disagree (unless there is a replacement system offered at some point). From my understanding, much of current CE sequencing load is onsie/twosie samples. Neither this or the Harmonia system will address these low throughput users, and the cost of maintaining a CE instrument is relatively high for most individual labs.

[krobison]

I would like to ask to the list what are the prospects of the GS Junior for transcriptomics:
probably not in the sense of digital gene expression analysis but more of the kind of research one sees in BMC Genomics, for cDNA discovery in species with no reference genome, which is specially important in plant genomics.

What are the current tools to filter/reduce the mRNA set to interesting medical/agricultural gene families? Considering the GS Junior will probably achieve ~100,000 reads per run, sample preparation will be essential to get the most of each run.

Anyone?

For a lot of species, it would seem a poor match for exploring novel transcriptomes for the reasons you mention -- too few reads. IF the PacBio instrument really can get long reads and IF there are ways to get access to the technology without blowing a few million $ (i.e. accessible through core labs or contract providers), then this would seem to be a much more promising option.

As for sample prep, one approach would be hybridization -- if you can pick your probes to very conserved regions and/or find somewhat less stringent conditions. Targeted mRNA-Seq does enrich for genes of interest & preserves significant expression information.

http://www.ncbi.nlm.nih.gov/pubmed/19835606

[Nomijill]

What is the expected cost per base from Junior?

[AKroxy]

The sales rep who I spoke to today about GS junior said 35 million bases per run (100,000 reads, 400bp long). She said the chemistry would cost about $1000/run + $200 for library construction.
So I guess this would be roughly 3 cents per base?

I'm wondering if she was low-balling the cost, or omitting some costs (like the picotiter plate) though, since she said the FLX chemistry was $5000/run including library construction and emPCR. I don't know of any core labs that charge less than ~$8000/run, not including library construction.

[pmiguel]

The sales rep who I spoke to today about GS junior said 35 million bases per run (100,000 reads, 400bp long). She said the chemistry would cost about $1000/run + $200 for library construction.
So I guess this would be roughly 3 cents per base?

I'm wondering if she was low-balling the cost, or omitting some costs (like the picotiter plate) though, since she said the FLX chemistry was $5000/run including library construction and emPCR. I don't know of any core labs that charge less than ~$8000/run, not including library construction.

About the 454 jr (or whatever it is called):
Yes, there is the cost of the instrument itself -- so, possibly depreciation on that must be considered. Then, after the first year, a service contract costing roughly 10% of the price of the instrument. If the instrument is like the GS-FLX, if you don't do a run during a week, you will need to do a maintenance run to sterilize the instrument tubing against biofilms. So, even without depreciation you are probably looking at $10K/year (after the first) without doing any runs. That is your fixed cost.

Then there is the cost of the staff to run the instrument. If you have a 454 jr. in you lab, you bear that cost. That is more difficult to assess because it might get spread among the staff in your lab, or focused on a single person. Either way, in the unlikely event that person otherwise did nothing but sit around watching Youtube videos all day--then staff costs you nothing. More likely, the person (or people) successfully running this instrument would be your most productive staff--who could be doing other experiments otherwise. So there is an opportunity cost to your staffing to add to the cost of ownership.

This is probably not an instrument you can throw into a common room and share among all the labs in a department--not unless you have some administrative structure around it to deal with scheduling, training and preventing someone not trained from attempting to use the instrument and breaking it.

35 million bases is about what you would get from one lane of a 16 lane gasketed PTP on the GS-FLX. Just to give you an idea of the capacity of the two instruments. So, using the figures you provide, the 454jr would have reagent costs/base of sequence roughly 3x those of the GS-FLX.

About the $5000/run reagent costs vs. the $8000+/run of the GS-FLX:

A core will almost certainly bear the same costs as I outline above -- service contract (approaching $1000/week), staffing costs and some will include depreciation on the instrument itself. (Someone has to pay for new instrumentation.) Also some of the runs during a year will fail and need to be repeated. Then there is the cost of titrations and other QC steps (fluorimetry, spectrophotometry, bioanalyzer runs, etc.)

And how about utilities and rent? If you are at a University you might think the University takes care of that. Not really. That is what "overhead" is. In our case it comes to an extra $54 on every $100 we bill out. Local faculty don't notice it since it budgeted separately for grants as "F&A" (facilities and administration). Some University Cores don't seem to have to charge their customers full overhead. But for us, and most other cores, I would think, it is there.

Then there is the matter of informatics. Realistically, there is a minimum amount you can get away with and still be providing a service labs will want. That minimum is probably supplying .sff or .fasta files to the investigator. For Titanium chemistry, that will consume 80 hours/run on the console computer. Much faster, of course, if you have access to a linux cluster where you can offload the image files and do processing there. But is that enough?

Even provided fasta file data, a naive lab faces a major learning curve to be able to do something useful with what we would consider a modest amount of sequence (35 million bases). At some cores (including Purdue) you will get first pass informatics -- assembly/clustering, blast searches, GO term annotation, a blast server hosting your data, etc.

From the core's point of view this is problematic. It costs more money to add an extra layer of informatics to your operations. And, from your comments above, one can see that many potential users base their decision on which core to use on cost/raw base or cost/run. But does that convert directly to cost/manuscript accepted or cost/degree or tenure awarded? Probably not. But we are trained from birth to consider "price" to be an entity in itself. Measuring "outcome"/dollar it far more difficult. So the only justification I can think of for a core to take on the extra cost for an extra layer of informatics would be that if you give your users something they can readily turn into a useful outcome quickly, they might be back for more in a shorter time frame.

So, count yourself lucky that you can find a core to do an $8000 GS-FLX run. They are either running a very tight ship or are being subsidized somehow. (That is, someone else is paying some of your costs.)

--
Phillip

[Enoch Root]

True Words

pmiguel thanks for your true words....cost per base is so much more than just chemistry...=)
Here is a another aspect, at least in Germany, half a picoPlate currently costs 4500 €, including everything from DNA to .sff and .fasta. I see hardly any reason to have a sequencer in ones own lab unless it really runs constantly.
If you have non standard questions to address you could even try to make a project together with the sequencing-company and get funding....