BioMicroCenter:Illumina Sequencing

From OpenWetWare
Revision as of 19:41, 25 May 2012 by Stuart S. Levine (talk | contribs) (Genome Analyzer IIx)
Jump to: navigation, search



The MIT BioMicro Center has five high-throughput Illumina sequencers, including a HiSeq 2000, three Genome Analyzers and one MiSeq. We support a wide variety of applications, such as ChIP-Seq, miRNA sequencing and RNA-seq. Each lane can potentially accomodate dozens of barcoded samples (depending on sequence complexity and desired coverage). Read lengths vary, depending on users, between 36nt and 150nt per end.
All questions about Illumina Sequencing can be directed to Kevin Thai at

Illumina Massively Parallel Sequencing

Cbot left.jpg
Illumina sequencing works by binding randomly fragmented DNA to an optical flowcell. Fragments are sequenced by sequentially incorporating and imaging fluorescently labeled nucleotides in a “Sequencing-By-Synthesis” reaction. Illumina recently rolled out its TruSeq v3 reagent kits, improving read quality and reducing GC bias at high cluster densities. For an in-depth overview of the Illumina sequencing chemistry, please refer to Kirchner et al 2009.


HiSeq 2000

Hiseq 2000.jpg
The Illumina HiSeq 2000 is the workhorse of BMC's Illumina fleet and is optimized for maximum yield and the lowest price per basepair. Each lane on the HiSeq is typically produces between 160 and 220 million reads passing our quality filter (for high quality libraries). HiSeq flowcells have 8 lanes, one of which is committed to a control sample that is used for base normalization (lane 1). Read lengths on the HiSeq very between 40 and 100nt per side and nearly all flowcells use barcoding to run multiple samples in each lane.

In order to optimize work flow and keep costs under control, only full flowcells are run. Since all 8 lanes of the flowcell must be run at equal lengths, submissions of single lanes must be grouped with other similar read lengths. This means that some read lengths move through our queue faster then others because more samples of that length are submitted to the BioMicro Center for sequencing. 40nt single end (SE) samples are by far the most common and move through the queue rapidly followed by short paired end (40+40) runs. Many lengths are very unusual (eg. 100nt single end) and can wait months for sequencing. We strongly recommend moving samples with unusual read requirements to one of the other platforms. If you have questions about this (or any other aspect of sequencing) please do not hesitate to contact us.

The HiSeq2000 is ideal for:

  • High numbers of multiplexed samples
  • De novo sequencing
  • SNP detection
  • ChIPseq
  • Bisulfite sequencing
  • RNAseq
  • Exome sequencing

The HiSeq2000 was donated to the BioMicro Center by Drs. Penny Chisholm and Chris Burge.


BMC miseq.png
The MiSeq is the newest sequencer in the BioMicro Center. Unlike the HiSeq, the MiSeq is optimized for speed. The MiSeq has a single lane that can produce approximately 5 million reads passing filter. The MiSeq does *not* have a control lane so having good base balance is critical for runs on this sequencer. Amplicons, such as 16S, can be run on the sequencer but should be constructed to have phased sequencing. Highly unbalanced libraries, such as RRBS, should not be run on the MiSeq.

The strength of the MiSeq is its speed and read length. The MiSeq is able to sequence 12nt/hour which allows it to complete a 150+150nt paired end read, from cluster to fastq files, in a little over a day. This compares to 2-3 weeks of sequencing on the HiSeq. Because the chemistry is on the flowcell for less time, error rates are much lower for the MiSeq then the HiSeq. New kits should push read length even longer, with 250+250PE kits coming soon and the Broad Institute having reported 400+400PE runs.

The MiSeq is ideal for:

  • Small genome resequencing
  • Targeted resequencing
  • Metagenomics
  • smRNA
  • barcode sequencing

The HiSeq2000 was donated to the BioMicro Center by Dr. Chris Love.

Genome Analyzer IIx

The Genome Analyzer II (GAII) are the oldest sequencers in the BioMicro Center and remain the most flexible. The newer generations of Illumina sequencers have been designed with increasing focus on clinical applications and have removed some of the "hands on" aspects of the older GAIIs. The GAIIs remain the only sequencers where the actual images of the flowcell can be reprocessed for example. The GAII/IIx can produce 20-40m reads per lane passing filter and typically runs read lengths of 36-150nt per side.

With the addition of the MiSeq, we have reworked how we are processing GAII flowcells. We have been able to create partial flowcells on the GAII by altering recipes. This has allowed us to move from a model like the HiSeq where we need a full flowcell before we run to a model where we can run as soon as the samples pass quality control, more like the MiSeq. However, unlike the MiSeq, we can run multiple lanes at once. Some critical caveats: First, these methods are not supported by Illumina so we cannot offer to replace failed runs. Second, unlike the HiSeq, the PhiX lane is *not* included. You must choose to sequence a lane of PhiX if you want to do control normalization. Finally, this service is completely "a la carte" so the pricing schema is quite different.

# of Lanes cycles per day cycles per kit
8 24 45
4 36 72
2 48 120
1 72 180

Using fewer lanes on each flowcell has allowed us to decrease the cycle time by not imaging all the lanes. In a typical 8 lane run, 20 minutes is spent doing chemistry followed by 40 minutes of imaging (each lane takes ~5 minutes to image). Therefore, a 2 lane flowcell runs twice as fast as an 8 lane flowcell. Also, since the chemistry is not running in to all of the lanes, each sequencing kit can go to a longer read length. The relationships are summarized in the chart on the left. Pricing is set on the number of lanes you are using, the number of days you are running the GAII, and the number of sequencing kits you are using. For example, if you wanted to run a 96+96 PE flowcell using 2 lanes, the cost would be the initial cost for the 2 lane PE flowcell plus an additional 3 days (one day is included in the original price) plus a second sequencing kit. That kit would not be completely used up (you would have an extra 48nt left that would be thrown away).

The GAII/GAIIx is ideal for:

  • Unusual read lengths
  • Protocol Prototyping
  • Non-standard assays such as HITS-FLIP

 This GAII pricing model is an experimental model and is subject to change 

The Genome Analyzer IIs were donated to the BioMicro Center by Drs. Penny Chisholm, Chris Burge, Ernest Fraenkel and the Dept of Biology with contributions from many others

Platform Comparison

SPEC HiSeq2000 GAII/IIx MiSeq
Machine Names FonZie Ryland
# reads / lane 160-220m 20-40m 4-6m
# lanes coprocessed 7+PhiX 1 to 8 1
nt / day 24 24-72 288


Illumina currently provides reagents and support for four major sequencing applications:

The following application has been published but does not yet have a kit from Illumina:

  • Genotyping: Protocols are being developed for detection of SNPs, chromosomal rearrangements and other genotyping applications.

Sample Preparation

Illumina sequencing requires the input of libraries which have been properly fragmented, ligated to specific adapters, and, in the case of RNA inputs, converted into complementary DNA. The BioMicro Center accepts fully prepared libraries from users and also offers a variety of sample preparation services for different applications.

Information is also available about multiplexing.


Quality control is very important for optimizing the number of reads and the quality of data produced. We run Bioanalyzer and RT-PCR for all submitted cDNA libraries for Illumina sequencing. For more information on QC methods and protocols please visit the Sequencing Quality Control page.

Data Analysis

Each flowcell lane flowcell should produce between 10 and 120 million DNA fragments as of September 2011. Understanding this data often requires a significant investment in informatics and many applications require entirely different interpretations of the data. As part of our sequencing service we provide many of the early steps of bioinformatics for different applications. Further data processing can be arranged on a collaborative basis as resources are available. For more information, check out the links below:


Protocols for all supported technologies can be found here.

Pricing and Priority

Full pricing information is available on our price list.

Priority for Illumina sequencing is currently given to labs associated with the BioMicro Center Core departments. We are able offer our services to other MIT and non-MIT users as space allows.

Turnaround Time

Each Genome Analyzer processes 8 samples per run, or 7 samples plus a control. Multiplexed pools of samples count as one for this purpose. The control is typically used to improve sequence quality. Full flowcells can usually be run within two weeks of submission. Partial submissions of less than eight samples (or 7 with control) are put into a project queue, where they join existing samples or await others before processing. Wait times for partial submissions vary depending on demand from other users. Once processing begins, approximately six days are required for clustering, sequencing, and data analysis for a 36-base-pair read. Longer reads add approximately one hour per additional base-pair including a second read and the barcode read, if either are present.

MIT Core Collaboration

Because of the layout of Illumina flowcells, samples must be run in batches of 7 lanes (a pool of multiplexed samples counts as one lane). In order to ensure quick throughput, we have established a collaboration that allows us to move partial flowcells between the various centers at MIT. For users with less then 4 samples, their samples may be moved between the BioMicro Center, the Whitehead Institute Center for Genome Technologies and the Koch Institute Biopolymer Center. Samples will be moved only to fill out runs or to expedite processing. The Centers are committed to working together to maintain consistent quality between the different cores, so you should see no difference whether your samples are run in BioMicro or at one of our sister centers. Transfers are only available for members of the MIT community.

View current samples queuing for Illumina

Initial page written by Summeet Gupta at the WI-CGT