BioMicroCenter:Illumina Sequencing: Difference between revisions

Revision as of 14:11, 25 March 2015

The MIT BioMicro Center has five high-throughput Illumina sequencers, including a HiSeq 2000, two Genome Analyzers, one NextSeq 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 325nt per end.
All questions about Illumina Sequencing can be directed to Leigh Manley or Michael Gravina.

Illumina Massively Parallel Sequencing

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. The BioMicro Center uses Illumina's 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.

Platforms

HiSeq 2000

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 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 should instead be submitted for the NextSeq unless you have a full flowcell. 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 HiSeq2000s were donated to the BioMicro Center by Drs. Penny Chisholm and Chris Burge and HHMI

MiSeq

The MiSeq is a newer sequencer in the BioMicro Center. Unlike the HiSeq, the MiSeq is optimized for speed. The MiSeq has a single lane that can produce up to 25 million reads passing filter (ideal cases). 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 14nt/hour which allows it to complete a 150+150nt paired end read, from cluster to fastq files, in a little less than 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. MiSeq runs are available in 50, 150*, 300, 500 and 600*nt flavors. (*) - v3 kits can reach 25m reads. Other kits can only reach 15m

The 50 cycle kit can accommodate up to 60bp read length (single-end or 30+30 paired-end). The 300 cycle kit can accommodate up to 350bp read length, while the 500 cycle kit can accommodate up to 520bp read length. New kits should push read length even longer, with the Broad Institute having reported 400+400PE runs.

The MiSeq is ideal for:

Small genome resequencing
Targeted resequencing
Metagenomics
smRNA
barcode sequencing

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

NextSeq

The NextSeq is the newest sequencer in the BioMicro Center. The NextSeq can be thought of as a MiSeq on steroids. Optimized for speed and yield, the NextSeq has a single lane that can produce up to 500 million reads passing filter (ideal cases). This yield does come at a slightly lower quality, and while most Illumina machines operate well above their specifications, the NextSeq has less margin. Like the Miseq, the NextSeq does *not* have a control lane, so having good base balance is critical for runs on this sequencer. In addition, the NextSeq chemistry only uses 2 fluor (instead of 4) which can complicate some experimental designs. Amplicons, such as 16S, have not yet been tested on the sequencer and may fail. Highly unbalanced libraries, such as RRBS, should not be run on the MiSeq.

The strength of the NextSeq is its speed and read length coupled with yield. The NextSeq is able to sequence ~10nt/hour which allows it to complete a 150+150nt paired end read, from cluster to fastq files, in two days. Because the chemistry is new, error rates are expected to be higher on the NextSeq than the MiSeq or the HiSeq. Kit sizes for Nextseq are 75, 150, and 300 nt. Currently, the BMC only stocks "High Output" flowcells. At this time, we discourage use of the NextSeq for 50nt reads - those are better suited to the HiSeq2000s - and we will prioritize other read lengths before 50nt runs.

The NextSeq is ideal for:

Whole genome sequencing
Splicing analysis in RNAseq
Metagenomics

Preliminary Data on NextSeq Runs (March 25, 2015):

We are closely monitoring the quality of NextSeq runs. So far we have found a decrease in quality as the number of reads increases. However, with the newly released v2 chemistry for Nextseq, we see a marked increase in quality (orange is v2, black is v1). For more information on NextSeq quality email Leigh at manleyL@mit.edu.

The NextSeq was donated to the BioMicro Center by Dr. Doug Lauffenburger, Dr. Penny Chisholm, the Dept of Biology and the Koch Institute.

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	42
4	36	66/33*
2	48	106/54*
1	72	140/81*
*Second number pertains to reads greater than 40 nt.

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 75+75 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 two additional sequencing kits. The last kit would not be completely used up (you would have an extra 18nt 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	NextSeq
Machine Names	SamAdams JackDaniels	Ryland Boris Preston	MiAmore	(TBD)
# reads / lane	160-220m	20-40m	12m-25m	300m-500m
# lanes coprocessed	7+PhiX	1 to 8	1	1
nt / day	18	24-72	288	150
Max Read Length	100+100	80+80 in lane by lane	325+300	150+150

Additional Information

Submission guidelines

Submission guidelines can be found on our FAQ.

Pricing and Priority

Full pricing information is available on our price list.

Priority for Illumina sequencing is given to labs associated with the BioMicro Center Core departments on a first-come first-served basis. Users requiring expedited services should contact Stuart Levine. We are able offer our services to other MIT and non-MIT users as space allows. A full description of priority and queue time expectations can be found on our FAQ.

Library Preparation

Illumina sequencing requires the input of libraries with inserts between 10 and 1000bp in length and have specific adapters attached to the 5' and 3' ends. The BioMicro Center accepts custom samples of all types provided the user also submits sequencing primers (though we do not assume responsibility if the samples fail). Samples submitted for Illumina sequencing should be at ~10ng/ul and the user should provide at least 10λ of samples. This is an ideal situation but we do have protocols available to help users with much less concentrated samples. Please submit your sample along with a completed Illumina sequencing form.

In addition to accepting finished libraries, the Biomicro Center supports a number of different sample preparation methodologies for different applications including RNAseq, ChIPseq and genome sequencing. All samples prepped in the BioMicro Center are barcoded for multiplexing.

Quality Control

The BioMicro Center undertakes a number of quality assurance methods to ensure that we produce high quality data for you. All samples submitted for Illumina sequencing are checked for size distribution, presence of proper 5' and 3' adapters, and actual concentration using the Agilent Bioanalyzer and qPCR. For more information on library quality can be found on the Sequencing Quality Control page.

We will skip pre-sequencing QC if the user supplies us with concentration and average fragment length information for each sample submitted. However, different labs often vary substantially in their quantifications and our methods are optimized for our own instruments and operators. We cannot guarantee optimal data output and quality for samples which are quantified outside of the BioMicro Center.

Additional quality metrics are done during all sequencing runs as part of the standard Illumina process. All samples are spiked with ~0.5% of the bacteriophage ΦX174. The ΦX library is primed off the standard Illumina sequencing primers and is used to both ensure the quality of the reagents used in the run and to measure the background sequencing error rates. ΦX reads will not be detected on non-standard libraries using custom priming.

Finally, several additional quality metrics are included in the automated analysis pipeline currently under active development in the Center. These include standard metrics of base composition, GC%, library complexity and overrepresented reads that are in the TagCount and Fastqc files. In addition, we now evaluate libraries for contamination from common laboratory species (human, mouse, yeast and E.coli). More information can be found on the Flowcell data guide page.

Pooling considerations

When determining how many samples should be combined together in a single lane, the following equations are useful:

# of lanes = (genomesize x coverage x #samples) / (#readsperlane x readlength x ends)
#samplesperlane = (#readsperlane x readlength x ends) / (genomesize x coverage)

where,

# of lanes is the total number of lanes that are required to achieve the specified coverage given the other variables
#samplesperlane is the total number of samples that can be combined into a single lane to achieve the specified coverage given the other variables
genomesize is the size, in nt, of the library to be sequenced
coverage is the desired multiplicity of coverage for the library
#samples is the number of samples needing to be sequenced
#readsperlane is the number of reads produced by a lane on the sequencer. (See "Platform Comparison" table above for the typical outputs from each platform.)
readlength is the length, in nt, of each separate read of the run
ends is the number of insert reads for the run. For single-end, it is 1, and for paired-end, it is 2.

Custom Primers

Many non-standard Illumina protocols require custom sequencing primers. The design of these oligos is critical for the success of the experiment and we have observed several experimental failures due to improper oligo design. There are a few critical parameters in oligo design.

First, the oligo must only occur once in each sequence. Multiple binding will result in low quality reads.
On reverse or index reads, we cannot separate the oligos by lane and so the construct must be compatible with having a cocktail of standard Illumina oligonucleotides in the mix.
The Tm of the oligo must match the Tm of the sequencing primer they are designed to replace. Being even a couple degrees below the Tm can result in experimental failure. Any online Tm calculator can be used. The standard Illumina sequences are:
Forward read: 5'ACACTCTTTCCCTACACGACGCTCTTCCGATCT
Reverse read: 5'CGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCT
Multiplexing Index Read: 5'GATCGGAAGAGCACACGTCTGAACTCCAGTCAC

Custom oligonucleotides should submitted at 100 µM with at least 20 µL provided. At time of submission, please take the time to diagram your primer design for the sequencing technician who can verify its compatibility and help to avoid any unforseen issues.

Data Analysis

Illumina sequencing at the BioMicro Center includes basic informatic analysis of the data. These steps include:

Image analysis to locate clusters
Basecalling
Demultiplexing of lanes
Alignment of sequences to a reference genome
Quality control
Delivery of the data to a user accessible folder

All of these steps are run by our automated analysis pipelinecurrently in active development. For users requiring further analysis, we have a staff of |bioinformaticists that can assist you in analyzing your data.

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 and the Whitehead Institute Center for Genome Technologies. 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.

OLDE LINKS

ChIP Seq
Small RNA Sequencing
Genome sequencing and resequencing
Illumina pipeline - How it works!
Illumina pipeline output formats
Bioinformatic Consulting
Protocols for all supported technologies can be found here