The good strategy can be running two techniques: Illumina as a working horse (eg one lane for a start), which is expected to give scaffolds cut into pieces by repeatable elements. Then PacBio hoping the assembled pieces will connect the scaffolds. This can eliminate artifacts of both technologies.

SMRT cells are not that expensive - you can add them until you are happy with the assembly - as the efficiency is hard to predict.

The real benefit of a PacBio machine over others is the extremely long reads. This makes it possible to blow through repetitive regions of the genome and still know where you are located. That being said, for your application (de novo assembly of a microbial genome) I think I would suggest either an Ion Torrent from Life Technologies (Cheapest WGS on the market and suited to small genomes) or if you really want to spend a bunch of money, buy the new NextSeq 500 from Ilumina.

Here is a link to some desktop sequencer specifications side by side:

http://nextgenseek.com/2012/08/comparing-price-and-tech-specs-of-illumina-miseq-ion-torrent-pgm-454-gs-junior-and-pacbio-rs/

Hope this helps.

This paper from last year says that you can do a whole genome assembly with as few as 3 SMRT cells:

Chin C-S, Alexander DH, Marks P, Klammer AA, Drake J, Heiner C, et al. Nonhybrid, finished microbial genome assemblies from long-read SMRT sequencing data. Nat Meth. 2013;10(6):563-9.

http://www.nature.com/nmeth/journal/v10/n6/abs/nmeth.2474.html

Price wise I can only tell you that in our inhouse sequencing facility we pay 2500 Euro per SMRT cell.

Hope this helps.

The good strategy can be running two techniques: Illumina as a working horse (eg one lane for a start), which is expected to give scaffolds cut into pieces by repeatable elements. Then PacBio hoping the assembled pieces will connect the scaffolds. This can eliminate artifacts of both technologies.

SMRT cells are not that expensive - you can add them until you are happy with the assembly - as the efficiency is hard to predict.

Ion Torrent is the wrong technology for de novo assembly. It suffers terribly from homopolymer run errors, it's reads aren't that long and you don't get that much data per run. I'm not quite sure what Ion Torrent is good for to be honest.

Cheapness is not a selling point when accuracy is sacrificed.

I'd say PacBio is perfect for microbial genome assembly. There's a great blogpost here which gives real world experience of applying PacBio to microbial genome assembly:

http://pathogenomics.bham.ac.uk/blog/2014/02/an-outsiders-guide-to-bacterial-genome-sequencing-on-the-pacific-biosciences-rs/

Although I agree the Ion Torrent is not really useful for human genome sequencing or even mouse genomes for that matter, Courtney is trying to sequencing an ~25Mb microbial genome. That is tiny in the sequencing world and I am certain the the read lengths of ~400bp are more than adequate for de novo assembly given that the 318 chip will produce more than enough data to assemble a 25Mb genome at 600 Mb–2.0 Gb of sequence data produced in ~5hrs.

I'm not saying it is good for EVERYTHING, but I don't agree that it is not good for ANYTHING. Besides, who can afford a PacBio machine? I only know one person who even has one of these. Almost any lab can afford an Ion Torrent (less than a good microscope).

There's a lot more to de novo assembly than raw numbers alone. You need high quality data to get a good de novo assembly. This paper kind of sums it up nicely:

http://www.biomedcentral.com/1471-2164/14/675

"PacBio-only assembly is comparable to hybrid assembly and significantly superior to assemblies performed with short read only data"

There's no need to buy a PacBio there are plenty of centres that will do PacBio sequencing as a service. That works out much cheaper than buying any sequencing machine (except perhaps a minION).

So that brings us back to her original question about cost. How much is PacBio sequencing? I have no idea. I do know the Ion Torrent chips are around $300-$500 depending on where you buy and which chip. Library costs may be around $1500-$2500 depending on if you do it yourself of have a core do it for you.

The blog I posted above talks about £350-£400 (US$580-US$670) per library + SMRT cell. There's a link to an interactive cost estimator at Duke (https://dugsim.net/estimate_cost) and there it quotes $704 ($408 library, $296 SMRT cell) for one library and SMRT cell.

Given those costs you could run 4-8 SMRT cells for each library for the same price as one 318 chip. That would give you at least as much data as the Ion Torrent.

Wow. That is a lot cheaper for PacBio than I thought. Good info.

We have had some experience with PacBio in our labs. We found PacBio compared to Illumina and 454 had higher error rates unless you can provide high quality and high quantity raw materials. It works best when used in hybrid, ie MiSeq and PacBio.

I have the same experience than Jenn Tan. We did a lot of Illumina sequencing (200x) without being able to really assemble the genome sequence of our favorite Eucalyptus species, even with highly experienced bioinformatician at work using a whole genome sequence of a related species (E grandis, see the open acess article just released by Nature : http://www.nature.com/nature/journal/vaop/ncurrent/full/nature13308.html. ). But then we started to redesign the sequencing strategy. Pac Bio RS2 seems to have been the good choice (we got the data some weeks ago and the new assembly is under progress) with dramatic cost decrease during the soe weeks we negotiated with the sequencing company (similar to the one cited by Christian Cole). Some tricky point about the concept of data amount per run. It seems that in fact the (or at least some) people operating the Pac Bio sequencer rather work with the amont of sequencing data produced rather than with a run (we initialy ordered a certain quantity of run with a fixed amont of data per un, but at the end they got the requested total amount of data (high quality date retainde after filtering) with half the number of runs (SMRT cells) expected... Could someone have further understanding of this ?

Our group opted out of it after we read and was told that PacBio has a very high error rate and does not work well by itself. It is a good option for highly repeated eukaryotic genomes for example where short Illumina reads are failing to spand. So I would suggest a combination of Illumina overlapping paired-end and mate-pair libraries assembled with Allpaths-LG and then PacBio for covering the missing bits.  

PacBio's chemistry has greatly improved, they are just releasing a new chemistry P6. With the PacBio you can now detect methylation too.  The pricing for PacBio has also dropped significantly.  The issue with PacBio though is always the quality of the DNA sample, you need high volume/conc, but if you can extract a good sample you could get very good results with good coverage.  The quality of the DNA can greatly impact the amount of data passing filter.

Well...as Already explained...PacBio Technology is the best at present to completely assemble the genome de novo.  Presently, the output on PacBio RSII with latest P6C4 chemistry is 500Mb-1Gb per cell. So for a 25 Mb Genome you will get 20-40X coverage in a single cell. So 3-4 SMRT cell would be enough to achieve 60-100X coverage.

The best strategy for you would be: 

1. Prepare a 20Kb library and sequence 4 SMRT Cell at  ~0.01nM on plate concentration.

2. If your, genome is haploid, then go for HGAP3 assembly pipeline or else for diploid genome FALCON can be used.

3. You will find near completely assembled genome.

Second one, sequencing on PacBio is presently little costlier than other platforms. But the positive thing is that you are going to have near finished assembled genome....what matters at last. Or either you will stuck with the short read sequences obtained thinking of what to do next....

I don't have much idea about costing scenario at your place but it will be better to plan de novo genome assembly using PacBio.

Hi, I am looking to send my DNA soil sample for sequencing. We are looking forward for sequencing nifH gene with IGK3/DVV primer set.

 I think the primer set IGK3/DVV may give an amplicon of ~394 bp, so will that be good to go with long-read sequencing technology or a shorter primer. Please let me know. 

 Thank you,

Hi Manoj,

For amplicon size you mentioned, it would be better to sequence on short read plateforms. It will be much cheaper and easier.

Think for the long read technology, if having amplicon > 1kb.

Thanks

Ratnesh

Hi Manoj,

I believes your gene of interest is of 1.8kb in length. Then why dont you plan to amplify and sequence the whole gene in your sample? Here long read technology will definetly help you.

Thanks