| ||per channel||per run
|raw nucleotides[hi 1]||~37,5Gb||~600Gb
|mappable with repeats||~90%
|mappable w/o repeats||~84% ###
forward: 5 days;
cluster restoring: 0,5 day;
reverse: 5 days
|throughput (mappable with repeats)||~49 Gb/day
|price[hi 2] (mappable with repeats)||~36 €/Gb
- ↑ for a 2x 101nt PE-run
- ↑ FC preparation and sequencing reagents only
[править] SOLiD 5500xl system / Applied Biosystems
| ||per slide||per run
|optimal beads density[s 1]||~15x108||~30x108
|raw nucleotides[s 2]||~180Gb||~360Gb
|mappable with repeats||~180GB (45%)
|mappable w/o repeats||~35% ###
|throughput (mappable with repeats)||~11 Gb/day
|price[s 3] (mappable with repeats)||~50€/Gb
- ↑ six deposition lanes per slide, ~250x106 beads for download per lane
- ↑ for a 2x60nt MP-run
- ↑ FC preparation and sequencing reagents only
[править] 454 (GS FLX Titanium Serie) / Roche
|throughput (filter-passed bases)||0.4-0.6Gb per run |
~1Gb per day
|run time||10 hours
|average length||400 bases
|accuracy||99.5% at 250 bases |
99% at 400 bases
|price per raw base||~13k€/Gb|
~30x higher than for Illumina/SOLiD
~10-30x lover than for capillary sequencing
|beads, no UV-lamps or lasers in sequencer
According to the throughput 454 is a next-generation sequencing system, but according to the price — not. A lot of applications are not practical for 454 because of high price per nucleotide.
[править] First generation (Sanger) sequencing
|throughput||50-100kb, 96 sequences per run
|accuracy||high quality bases - 99%: ~900bp|
very high quality bases - 99.9%: ~600bp
|price per raw base||~400k€/Gb
[править] Comparison of Illumina GAII and HiSeq2000 and SOLiD 5500xl systems
[править] General comparison (I>S)
Let’s suppose, that it is necessary to resequence human genome with a 30x coverage using "long-insertion" mate-paired procedure. This is about 30 x 3Gb = 90Gb of mappable nucleotides.
One MP-run on Illumina gives ~600Gb of raw nucleotides or ~540Gb (~90%) of mappable nucleotides.
Number of runs: 90Gb / 540Gb = 0.17 of a run, ~3 lanes of a flowcell
Project output: ~4.5x10^8 of 2x101nt reads with an error rate ~### errors per 100 of 101nt reads.
* library preparation: ~€100 on non-kit reagents
* flowcell preparation (per FC): €4127
* sequencing (per FC): €5616
* consumables (per FC): ~€45
TOTAL: (€4127 + €5616 + €45) x 0.33 + €100 => ~€3330
* library preparation: a bit more, than 1 week, let's say 10 days
* sequencing: 11 days per run with two flowcells
TOTAL: ~21 day
One MP-run on SOLiD gives ~180Gb of mappable nucleotides.
Number of runs: 90Gb / 180Gb = 0.5
The result is ~7.5x10^8 of 2x60b reads with an error rate ~### errors per 100 of 50bp reads.
* library preparation: ~€200 on home-made reagents
* flowcell preparation (per run): €3087
* sequencing (per run): €5943
* consumables (per run): ~€100
TOTAL: (€3087 + €5943 + €100) x 0.5 + €200 => ~€4765
* library preparation: a bit more, than 1 week, let's say 10 days
* sequencing: FC preparation 1 day, 16 days per run with two flowcells, 17 days total
- project time is practically the same;
- a bit more hand work for SOLiD, because of ePCR;
- price is 1.4 times cheaper for Illumina;
- error rate is ###.
[править] Which system is better (I>S)?
- Illumina is currently better in terms of throughput and price.
- all Illumina models were so far more "ripe", than SOLiD versions appearing at the same time. This means they were more stable in terms of hardware problems and produced usually more data than promised by the company, which was never the case with SOLiD.
- both platforms have same applications.
- if machine is planned to be used for "all possible applications" in a service center of an institute, Illumina is better, because it is more flexible (wider range of PE-libraries sizes; easier multiplexing; less hands-on time for flowcell preparation).
- however for small-scale and/or small throughput sequencing SOLiD 5500xl might be more flexible, since different types of sequencing runs may be performed on different lanes of the flowcell, and only part of lanes may be used in a run.
[править] Throughput (I>>S)
- throughput (mappable nucleotides per day) is higher for Illumina. SOLiD 5500xl has so far not reached the promised output. Some time ago the company was talking of using smaller beads, which could increase the throughput, but it is now unclear when this update occurs.
[править] Insertion size (I>S)
- for Illumina: on surface amplification works for <700bp fragments, which corresponds to ~600bp inserts. However, fragments in the range 300-600bp form heterogeneous clusters. They may be sequenced, but with lower efficiency, because cluster density should be lower.
- for SOLiD: "beads / emulsion PCR" works well for fragments <350bp, which corresponds to <300bp inserts.
- if starting material is a set of heterogeneous in length short DNA fragments (like, for example, in chromatin immunoprecipitation), more fragments would be suitable for sequencing on Illumina, than on SOLiD.
[править] Read length (RL) (I>>S)
- possible read length is about twice larger in Illumina (150nt) if compared with SOLiD (75nt). It seems that this ratio would not change in the near future. RL on HiSeq2000 is so far 100nt, but it seems the increase is just a metter of time.
- the difference between two platforms in achieved RL is not principal: ~50nt reads are long enough to be aligned to a genome; 150nt reads are still not enough to cover a repetitive region.
- long RL is not always necessary. Optimal RL is different for different applications.
Comparison of SOLiD and Illumina sequencing errors
[править] Quality (I∼S)
- #### this paragraph has to be updated: forward and reverse reads, possibility of 7th primer in SOLiD protocol which increases the quality.
NB! accuracy is not uniform along the read length. It is extremely high nearby with the RL-limit. Comparison of the systems gives different results on different read lengths.
- accuracy of SOLiD platform is higher, because each nucleotide twice in this ligation-based sequencing is analyzed. As a result, single-nucleotide substitution changes color of two dinucleotides. "Standing alone" color changes may be filtered out as sequencing errors after alignment.
- NB! alignment is absolutely necessary for filtering out of "standing alone" color changes in SOLiD data.
- comparison of error frequencies: "standing alone color mistakes SOLiD" >> "Illumina errors" > "filtered two-dinucleotide color changes SOLiD".
- coverage variance of the SOLiD is about two times higher than coverage variance for Illumina, which is not good for low-coverage analysis. Besides, high variance of coverage on SOLiD makes analysis of structural variations more difficult. Taken together, combination "accuracy + coverage" is about the same for SOLiD and Illumina.
[править] Perspectives (I>S)
- development strategies were so far quite similar for both companies: increasing of seq-elements density, increasing the imaging area, better chemistry and enzymes, acceleration of imaging, optimization of seq. kits composition, minimization of hands-on time, etc. Now Illumina platform is ahead, however SOLiD has some plans to improve (e.g. smaller beads).
- polymerization and ligation approaches: it seems, that polymerization provides a higher read length. However ligation-based technique is stable enough to survive. 50nt is enough for stable alignment fragments to the human genome. Both 75nt and 150nt are too short for jumping over repetitive sequences.
- From a technical point of view, both machines are combination of a liquid handling robot with a fluorescent scanner. In case, some particular sequencing approach would appear to be much better if compare with a competitor, both platforms might be adapted for the optimal procedure.
- #### distant plans of Illumina are to increase sequencing elements density by using a prestructured surface/ beads combination. Illumina has a large experience using beads for hybridization arrays, so, most probably they would succeed.
[править] Fragment libraries (I∼S)
- fragment libraries for both systems are DNA fragments with adaptors attached to both ends. Any procedure/kit developed for one system may be relatively easy adapted for the other.
- the same story for any single-read (or single read + barcode) library: fragment genomic libraries, microRNA libraries, etc.
- fragment libraries for Illumina may be prepared for larger range of insert sizes.
- minimal possible amount of starting material is about the same for both systems.
[править] PE-libraries <300bp (I>S)
- Illumina has developed a technology for changing orientation of clusters directly on the glass. It gives a possibility to sequence library fragments from both ends with about the same efficiency.
- ligation-based technology is able to sequence fragments in both directions. But now it works better in 3'→5' direction (up to 75b), than in the opposite 5'?3' direction (up to 25b); PE kits for SOLiD should appear on the market end of 2010.
[править] PE-libraries 300-600bp (Illumina only)
- "PE sequencing" is restricted by length on both Illumina and SOLiD platforms:
- MP-libraries can't be constructed from relatively short (<500bp) fragments, because they can't bend to form a circle.
As a result, 300-600bp PE-libraries may be constructed only on Illumina platform. This may be important for libraries with heterogeneous inserts, for example, ChIP-libraries.
[править] MP-libraries >600bp (I?S)
- large insert MP-libraries for the two platforms differ only by sequences of adaptors. Any protocol may be adapted for both systems. For example, MP-library construction protocol for Illumina presented on this site is based on SOLiD procedure.
[править] Flowcell preparation (I>>S)
Even with automation of beads preparation with EZ Bead System, ePCR remains time-consuming, contamination unsafe, hardly parallelized procedure.
|parameter||Illumina GA II and HiSeq2000||SOLiD 5500xl||comments
|time||+++||-||~3 times less for Illumina
|parallelisation||++||-||Beads for SOLiD can be prepared for max. 3 libraries a day
|hand-work||+++||+||more for SOLiD
|small-scale library prep||+++||--||Illumina: no restrictions|
SOLiD: it is impossible to prepare PCR emulsion for less than 1/2 of the lane because of the volume of ePCR bags (E10 has minimal volume)
|loading of several samples||++||++||Illumina: 16 independent samples may be loaded without any loss of throughput;|
SOLiD: 12 independent samples
|contamination safety||++||+||much better for Illumina procedure
|flowcell storage||+||####||days for Illumina, #### for SOLiD
[править] Multiplexing (I>S)
- 8 independent samples may be loaded on one flowcell without any loss of throughput (in separate channels);
- adaptors and/or PCR primers may be easily redesigned by insertion of "coding regions"; code might be included in the library both at ligation and final amplification steps.
- "multiplexing kit" is available for parallel analysis of 12 samples in one channel (96 analyses on one FC).
- 4 independent samples — loss of 1/4 of sequencing area; 8 samples — loss of 1/3 of sequencing area;
- it is very difficult to prepare PCR emulsion for less than 1/2 of the slide because of geometry of ULTRA TURRAX DT-20 tube;
- "multiplexing kit" is available for parallel analysis of 96 samples on one glass.
[править] Sequencing (I?S)
|parameter||Illumina GA II||SOLiD||comments
|reagent preparation||+++||++||Illumina: ready-to-load kit|
SOLiD: about 1 hour for preparation of solutions and stripes.
|flowcell installation||?||about the same hand-work
|hands-work||?||both platforms: one reagent loading for one-direction sequencing
[править] Bioinformatics (I>S)
- both systems require a strong bioinformatic support. It is impossible to use them without additional computer power and a skilled computer specialist. Bioinformatician should:
- install the software from the company;
- select and install software from the external sources;
- write own scripts;
- organize data storage and transfer.
- normal nucleotides from Illumina are a bit more convenient, than color space of SOLiD.
[править] In house development (libraries) (I>S)
- flanking sequences of Illumina libraries should correspond to the oligonucleotides immobilized on the flowcell. This is the only restriction in designing alternative adapters and PCR primers. Introduction of alternative sequencing primers is easy and straightforward. It is possible to use a different sequencing primer on each lane of a flowcell (for single read runs or for the first read of a PE run).
- it is problematic to modify common regions in SOLiD libraries. As for Illumina, one of the flanking regions should correspond to the primer immobilized on beads. It is impossible to change sequencing oligos without "cracking" the kit: apart from at least five different sequencing primers per read special "blocking" and "bridge" oligos are used during sequencing. ABI neither supplies information about structure of these oligos, nor takes orders for custom synthesis.
[править] In house development (flowcell) (I<S)
- SOLiD flowcell is a plain amino-modified glass slide. It is possible to customize bead deposition, etc. Practically nothing is possible to do inside Illumina's channels.
[править] In house development (sequencing) (I?S)
|parameter||Illumina GA II||SOLiD 3+||comments
|modification of the instrument programm||+++||-||Illumina: easy and clear both for image collection and liquid handling|
SOLiD: impossible[s 1]
|fluorescence excitation and detection||+||++||Illumina: uses two lasers, excitation wavelengthes are practically fixed|
SOLiD: UV lamp, in principle it is possible to change filters
- ↑ EcoP15I MP-libraries have 27b-long inserts, but it was possible to select only (20, 25, 30, 35, 40,...) sequencing cycles in SOLiD instrument control software. Changing of read length from 25 to 27 would be really helpful. We tried to push the company to modify the program, but without success. On Illumina it would have taken ~5min to change the number of cycles.
[править] Kits and consumables (convenience) (I?S)
During the last year, both companies have considerably optimized composition of kits and minimized hands-on time for preparing sequencing reagents. SOLiD library preparation kits appeared on the market.
Illumina and SOLiD library preparation kits:
- some reagents, mostly enzymes, should be dispensed by 2µl pipette, but are delivered in 1.7ml tubes: Klenow, Klenow exo(-), primers.
- side labels on some tubes completely hide the content of the tube;
- tube caps without labels: confusing.
- misleading names of reagents:
- two completely different reagents have similar names: "Bead Wash Buffer" and "Bind & Wash Buffer";
- "Low Salt Binding Buffer" instead of "Binding Buffer";
- quantities of viscous solutions (glycerol, ePCR components) should be expressed in terms of weight (not volume);
[править] Additional equipment (I?S)
both systems require:
- additional computing facilities for data storage and analysis;
- room with climate control (NB! SOLiD produces more heat);
- nebulizer is a bad choice for DNA shearing: an ultrasonic device should be ordered (e.g. Covaris S2);
- Hydroshear is obligatory for preparation of large-insertion libraries;
- Xstream pipette;
- should be enough number of 96-well PCR machines for ePCR.
[править] Ordering and delivery (I<<S)
- Illumina: delivery takes at least 3 weeks (Europe).
- SOLiD: normally 1-3 days from order to delivery. Excellent.
[править] Stability of the system (I>S)
- both systems are not completely stable. Sometimes (one per ~20) run fails for unknown reasons. Illumina is a little bit more stable than SOLiD in this respect.
- Illumina uses lasers, SOLiD — an UV lamp. Lasers are much more expensive, but UV-lamp has a much shorter operation life (about 0.5 year). Each UV lamp costs ~$600.
[править] Service and support (I<S)
- people from Applied Biosystems are ready to help in setting up of basic techniques, response very quickly in case of problems, help to analyze them and help to solve them. ABI has access to all running instruments, so sometimes they notice a problem sooner than a user.
- training courses for library preparation and technical assistants are available.
- reparation is fast, substitution for broken parts is delivered in a next day.
- substitution of kits for failed runs (if it was an instrument problem) is easy; for Illumina one has to spend a lot of time persuading tech support it was an instrument problem.
- but Illumina procedures are significantly easier and machine is a bit more stable.