Pathogens: Genes and Genomes

Yesterday, we launched a new version of the high-throughput sequencing map. I promised when it launched that when sufficient updates were received we would release instrument statistics. We have now received 214 updates from the community – thank you so much! So, without further ado, check out the statistics tables!

Some bottom-line numbers: there are (as of a minute ago) 669 machines listed in the database, spanning 250 sequencing facilities, an average of 2.7 machines per centre (a bit like a UK family having an average of 2.4 children!)

The take-home message which won’t surprise anyone is that Illumina are still dominating the market-place. 406 GA2s are accounted for, making up about 60% of our sample. This means they have shipped nearly four times as many sequencers as their two main rivals, Roche’s 454 and Life Technologies’ SOLiD.

Perhaps more interestingly is the battle for 2nd place. According to our sample Roche are just edging SOLiD with 118 machines to 110 but the competition is intense. This indicates that SOLiD have made up a huge amount of ground in the past year or two. Pretty good for a machine we wrote off as being the “worst of both worlds” last August. It is not clear to me how they have managed this impressive feat (although I have heard some rather uncharitable possibilities from people), but it is good news for Life Technologies that they have stayed in the game and are apparently gaining traction. The battle ain’t over yet.

About half the 454 and Solexas are in dedicated genome centres like the Broad, BGI and the Sanger Centre, but only about a third of SOLiDs are in dedicated genome centres, suggesting this platform has gained popularity in the smaller facilities. The largest SOLiD installation that we know about is at the University of Queensland with 11 machines.

There’s much less good news for Helicos – the Heliscope’s early entrance into the so-called “3rd generation” market appears to have gained virtually no traction in the market at all, with only 8 machines accounted for on the map.

Country-wise, the US still dominates the world with 46% of the machines in North America. We are pleased to see that the UK is well-represented with 95 machines and 29 facilities, although there is probably a bias as we hail from these isles.

China are the current world #3 in terms of sequencing machines but this is likely to change very soon when BGI start taking delivery of their HiSeq 2000 instruments, with 128 apparently on pre-order. Hopefully someone from BGI will come and update the map when they arrive.

And of course, we expect this map to have a few new faces in a years time with at least Pacific Biosystems and perhaps Ion Torrent bringing their products to market.

No doubt you can find other interesting nuggets in the stats, please feel free to share them in the comments.

Finally, a plea. If you see inaccurate or missing information on the map, please don’t complain, just get involved! This is a community resource that relies on your help. It’s really easy to add and update details on the map and if you have problems just get in touch.

One last caveat, I am not saying there are only 669 machines in the wild, the real number is conservatively likely to be 4- or 5-times this. We want to confront this problem head on and find out exactly what sampling we have got to help know how far the map has to go before it is complete.

Therefore we have set up the the great high-throughput sequencing serial number challenge. Taking inspiration from the German tank problem we wish to accurately estimate the numbers of machines shipped by reference to machine serial numbers. This was inspired by a recent thread on SeqAnswers.com. By allowing anonymous submissions we hope to collect a statistically useful set of serial numbers. Help us out, please visit the site and help us out by submitting your serial numbers!

A few months back I blogged about James Hadfield’s excellent Google Map for locating next-generation sequencing facilities. Some time has passed and the map has become very popular with over 100,000 views registered. Keen to create this into a really great community resource, James and I have been collaborating in our spare time on a revamped version. There are a couple of major improvements over the first version;

• you can now filter down to locate and compare sequencing platforms
• the placemarkers are now clustered to make the map faster and to make it easier to visualise hotspots of sequencing machines
• it is now easy to add new genomics platforms (including microarrays) as they become available
• you can easily add and edit facilities, which is subject to moderation to prevent spam

Perhaps most excitingly for stats-nerds, this is all database-driven so the long awaited league tables of sequencing platforms can be calculated automatically. However, mindful that the information could be made more accurate, I have decided to leave the league tables off until people have had a bit of time to check the map for errors, and add any missing facilities.

We’re really keen to gather feedback so either post a comment on the blog, or get in touch directly if you have ideas you want to discuss. We were also thinking about extending it for other non-genomics utilities, for example next-generation proteomics. If you are working in such a field and would be interested in collaborating, again get in contact.

Well, what are you waiting for? Go check out the map right now!

James Hadfield has created a very nice community curated map of all known next-generation sequencing facilities. Although perhaps not 100% accurate it gives a good feel for the relative adoption of 454, Solexa and ABI in the UK. I’ve summarised the table as of today (plus adding in Birmingham’s new 454 machine). Excluding the Sanger Centre which skews the result, it can be seen that Solexa is just in the lead with 12 machines, closely followed by 454 with 9 machines and with ABI SoLiD trailing with just 3.  Sanger are listed as having 40 Solexas (but curiously no 454s, but I know from a previous visit they have at least a couple) which obviously puts Illumina in clear first place in the UK. If you know of more machines, don’t tell me but instead post on this thread.

Apologies to my alma mater for leaving in the ‘Westfield’, force of habit, of course back in the day we just called it Barts & the London!

Update: Prof Neil Hall from Liverpool’s Centre for Genomic Research has written to say they now have two 454s installed! As the data is constantly changing, please refer to the original Google Map for the current scores on the doors.

David Dooling has uncovered an intriguing letter from Kevin McKernan, Senior Director of Scientific Operations at ABI sent to the House of Lords last November. The letter concerns the Sanger Centre’s decision to return their five ABI SoLiD instruments in order to concentrate on building Illumina (Solexa) capacity. This was of course a big blow to ABI – the Sanger Centre is Europe’s largest sequencing facility.

The explanation for sending the machines back was that the Illumina required less investment in molecular biology scale-up within the Sanger Centre’s existing infrastructure and wasn’t a reflection of the machines technical merits. Similar decisions were taken elsewhere, Wash-U also decided to focus on the Illumina platform. This letter criticises the Sanger’s decision accusing it of “taking a more historical approach” in their decision to build a homogenous Solexa platform and levels accusations of bias, noting that Sanger Centre staff have close ties with Solexa. Calling the approach historical is slightly ironic given that ABI used to be the only show in town and the Sanger had over a hundred ABI machines running during the HGP. The discovery of close ties are hardly surprising, Solexa is headquartered near Hinxton in Cambridgeshire and was part-funded by funding bodies such as the BBSRC.

Our reading of the situation was that Illumina have been successful because they were early to market, offered the best throughput and had a relatively simple sample preparation and data analysis pipeline. Solexa has a reputation for its decent sample preparation workflow due to its “walk away” Cluster Station machine. Solexa also produces easy to manipulate FASTQ read files which are read by all the short read mappers without the additional informatics distraction of SoLiD’s “colourspace”. Solexa has also found a way to scale rapidly from 1 gigbase per run and 36-base reads with its Genetic Analyzer 1. Now they are now routinely able to produce over 10 gigabases (and climbing) with 75-base reads and paired-ends to make analysis easier. 454 has also found a highly successful niche, particularly in bacterial applications and metagenomics, due to its longer reads, short run time and user-friendly software. SoLiD’s advantages were harder to define, sometimes characterised as “the worst of both worlds”, with a laborious sample preparation workflow, long running time, short reads and an unfamiliar output format.

I guess you cannot blame ABI for being upset that SoLiD has not been adopted as the de facto platform for short-read sequencing at many genome centres, but am not convinced that the decision represents any form of bias. It makes sense that a large sequencing centre would wish to scale-up a single technical solution when possible – each platform has a different workflow, the staff need different training, the molecular biology investment in equipment is different and the informatics needs different software. David Dooling entitled his post “sour grapes” and he may have a point.

However, ABI seem to be bouncing back, regardless of the centres which are SoLiD-free zones . Anecdotally, the people that have got them in the UK speak very favourably of them. There is a good chance we may be getting one at the University of Birmingham. The throughput is excellent and a recent report has demonstrated a single human genome in a single run. Clearly none of these technologies wish to sit on their laurels, and it would be worth continually reappraising their strengths and weaknesses as new hardware and software updates are released.

What are your experiences with this platform?