Benchmarking Differential Gene Expression Tools

In a recent study, Schurch et al., 2015 closely examine 9 differential gene expression (DGE) tools (baySeq , cuffdiff , DESeq , edgeR , limma , NOISeq , PoissonSeq , SAMSeq, DEGSeq) and rate their performance as a function of replicates in an RNA-Seq experiment. The group highlights edgeR and DESeq as the most widely used tools in the field and conclude that they along with limma perform the best in studies with high and low numbers of biological replicates. The study goes further, making the specific recommendation that experiments with greater than 12 replicates should use DESeq, while those with fewer than 12 replicates should use edgeR. As for the number of replicates needed, Schurch et al recommend at least 6 replicates/condition in an RNA-seq experiment, and up to 12 in studies where identifying the majority of differentially expressed genes is critical. 

With each technical replicate having only 0.8-2.8M reads, this paper and others (Rapaport et al., 2013) continue to suggest that more replicates in an RNA-seq experiment are preferred over simply increasing the number of sequencing reads. Several other papers, including differential expression profiling recommendations in our Sequencing Coverage Guide recommend at least 10M reads per sample, but do not make recommendations on the numbers of replicates needed. The read/sample number disparity is related to the relatively small and well annotated S. cerevisiae genome in this study and the more complex, multiple transcript isoforms in mammalian tissue. By highlighting studies that carefully examine the number of replicates that should be used, we hope to improve RNA-seq experimental design on Genohub

So why don’t researchers use an adequate number of replicates? 1) Sequencing cost, 2) Inexperience in differential gene expression analysis. We compare the costs between 6 and 12 replicates in yeast and human RNA-Seq experiments using 1 and 10M reads/sample to show that in many cases adding more replicates in an experiment can be affordable. 

 

6 replicates

12 replicates

Human (10M reads/sample)

$2,660

$4,470

Yeast (1M reads/sample)

$2,810

$4,470

*Prices are in USD and are inclusive of both sequencing and library prep cost. Click on prices in the table to see more project specific detail.

The table shows that the main factor in the price difference is related to library preparation costs. Sequencing on the Illumina Miseq or Hiseq at the listed sequencing depths do not play as significant a role in cost, due to the sequencing capacity of those instruments. 

To accurately determine the sequencing output required for your RNA-seq study, simply change the number of reads/sample in our interactive Project Page

 

References:

Evaluation of tools for differential gene expression analysis by RNA-seq on a 48 biological replicate experiment. Nicholas J. Schurch, Pieta Schofield, Marek Gierliński, Christian Cole, Alexander Sherstnev, Vijender Singh, Nicola Wrobel, Karim Gharbi, Gordon G. Simpson, Tom Owen-Hughes, Mark Blaxter, Geoffrey J. Barton

Comprehensive evaluation of differential gene expression analysis methods for RNA-seq data. Franck Rapaport, Raya Khanin, Yupu Liang, Mono Pirun, Azra Krek, Paul Zumbo,Christopher E Mason, Nicholas D Socci and Doron Betel

How Many Replicates are Sufficient for Differential Gene Expression?

In a nicely done, convincing 2 condition, 48 replicate RNA-Seq experiment, researchers from the University of Dundee aimed to answer a frequently asked question in the field and on Genohub.com, ‘How many replicates are necessary for differential gene expression (DGE)?’. In their study they examined three statistical models to see which best represented read-count distribution of genes from commonly used DGE tools.

Using the statistical power of 48 replicates they determined that inter-lane variability does not play a large role in DGE results. Assuming even loading and amplification their results showed a Poisson distribution of counts from individual genes. The authors also determined that read count distribution was consistent with a negative binomial model, an assumption in widely used tools such as edgR, DESeq, cuffdiff and baySeq. Performing a goodness-of-fit test for log-normal, negative binomial and normal distributions the authors demonstrated that inclusion of ‘bad replicates’ made results inconsistent with the statistical models they tested, complicating the interpretation of differential expression results. A bad replicate was defined as 1) one that poorly correlated with other replicates, 2) a replicate with atypical read counts, 3) one having non-uniform read depth profiles. 

So how many replicates are sufficient for differential gene expression? The authors sequenced 96 mRNA samples in 7 1×50 HiSesq lanes. The cost for this on Genohub today is $28k USD. When the authors removed 6-8 bad replicates from their pool of 48 samples, their data became consistent with a negative binomial distribution. Assuming experimental variability similar to the authors, this indicates at least 6 replicates in a DGE experiment is good practice. The cost for the preparation of 6 RNA-seq libraries and sequencing is $2,500 USD.  

Doing a literature search and observing client behavior on Genohub, we estimate ~80% of those studying DGE use 3 replicates in their experiments, which with dropouts and variation is unsatisfactory. A final point we’d like to make is the authors used ~11M, 1×50 reads/sample, which goes to show that with DGE, replicates can be more important than read depth. This is further discussed in our Coverage and Read Depth Guide

At Genohub, we help consult and design sequencing experiments with users. Determining the replicates needed for a study is a common question that needs to be answered. Unfortunately, too few studies examine these fundamental elements in sequencing design. We hope this article gets the recognition it deserves. 

Reference: 

Statistical models for RNA-seq data derived from a two-condition 48-replicate experiment Marek Gierliński, Christian Cole, Pietà Schofield, Nicholas J. Schurch, Alexander Sherstnev, Vijender Singh, Nicola Wrobel, Karim Gharbi, Gordon Simpson, Tom Owen-Hughes, Mark Blaxter, Geoffrey J. Barton

 

10 Sequencing Based Approaches for Interrogating the Epigenome

cytosine methylation

DNA methylation occurs when DNA methyltransferase transfers a methyl group from S-adenosyl-methionine to cytosine in CpG dinucleotides. The methylation of 5’ methyl cytosine (5mC) nucleotides is an important epigenetic change that regulates gene activity and impacts several cellular processes including differentiation, transcriptional control and chromatin remodeling.

Genome wide analysis of 5mC, histone modifications and DNA accessibility are possible with next generation sequencing approaches and provide unique insight to complex phenotypes where primary genomic sequence is not sufficient.

Methods for methyl DNA sequencing can be broken down into three global approaches:

1) bisulfite sequencing

2) restriction enzyme based sequencing

3) targeted enrichment of methyl sites

We’ve outlined several library preparation techniques under each category.

1) Bisulfite Sequencing

Bisulfite-seq (1, 2) is a well-established protocol that provides single base resolution of methylated cytosine in the genome. Genomic DNA is bisulfite treated, deaminating un-methylated cytosines to uracils, which are later converted to thymidines. Methylated cytosines are protected from deamination, allowing researchers to identify methylation sites by comparing the sequence of bisulfite and non-bisulfite treated samples.

1-Shotgun bisulphite sequencing of the Arabidopsis genome reveals DNA methylation patterning

2-Highly integrated single-base resolution maps of the epigenome in Arabidopsis

2) Post Bisulfite Adapter Tagging (PBAT)

To avoid loss of template during bisulfite treatment, with PBAT (3) bisulfite treatment follows adapter ligation (tagging) and two rounds of random primer extension.

3- Amplification-free whole-genome bisulfite sequencing by post-bisulfite adaptor tagging.

3) Reduced Representation Bisulfite Sequencing (RRBS)

RRBS (4) is a method aimed at targeting sequencing coverage toward CpG islands or regions of the genome with dense CpG methylation. Sample is digested with one more restriction enzymes and then is treated with bisulfite prior to sequencing. This method offers single nucleotide methylation

4- Reduced representation bisulfite sequencing for comparative high-resolution DNA methylation analysis

4) Oxidative Bisulfite Sequencing (oxBS-Seq)

5-hydroxymethylcytosine (5’hmC), an intermediate of the demethylation of 5-methylcytosine (5’mC) to cytosine cannot be distinguished using the bisulfite-seq approach. With oxBS-Seq (5), 5’hmC is oxidized, causing a deamination to uracil, while leaving 5’mC. Sequencing of both treated and untreated samples allows for single base resolution of 5’hmC and 5’mC modifications.

5-Quantitative sequencing of 5-methylcytosine and 5-hydroxymethylcytosine at single-base resolution

5) TET-Assisted Bisulfite Sequencing (TAB-Seq)

TAB-Seq (6) utilizes glucose moieties to interact with 5’hmC protecting it from TET protein oxidation. 5’mC and non-methylated cytosines are deaminated to uracil and sequenced as thymidines, allowing for the specific identification of 5’hmC.

6- Base-resolution analysis of 5-hydroxymethylcytosine in the Mammalian genome

6) Methylation Sensitive Restriction Enzyme Sequencing (MRE-Seq)

MRE-Seq (7) utilizes a combination of methyl sensitive and insensitive restriction enzymes to identify regions of CpG methylation status.

7- Genome-scale DNA methylation analysis

7) HpaII tiny fragment-Enrichment by Ligation-mediated PCR (HELP-Seq)

HELP-Seq (8) allows for intragenomic profiling and intergenomic comparisons of cytosine methylation by using HpaII and its methylation insensitive isoschizomer MSPI.

8- Comparative isoschizomer profiling of cytosine methylation: the HELP assay

8) Methylated DNA Immunoprecipitation Sequencing (MeDIP)

MeDIP (9) is a technique based on affinity enrichment of methylated DNA using either antibodies or other protein capable of binding methylated DNA. This technique pulls down heavily methylated regions of the genome, such as CpG islands. It does not offer single nucleotide resolution.

9- Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells

9) Methyl Binding Domain Capture (MBD-CAP)

MBD-CAP (10) uses methyl DNA binding proteins MeCP2, MBD1-2 and MBD3LI to immunoprecipitate methylated DNA. Similar to MeDIP, this approach pulls down regions that are heavily methylated and does not offer single nucleotide methylation resolution.

10- High-resolution mapping of DNA hypermethylation and hypomethylation in lung cancer

10) Probe Based Targeted Enrichment

Methyl-Seq Targeted enrichment – involves the use of synthetic, biotinylated oligonucleotides designed to CpG islands, shores, gene promoters and differentially methylated regions (DMRs). Kits are commercially available through Agilent and Roche Nimblegen.  

Finally, it’s worth mentioning Single Molecule Real Time (SMRT) DNA Sequencing. SMRT sequencing by Pacific Biosciences uses the kinetics of base incorporation to allow for direct detection of methylated cytosines. Unlike any of the protocols mentioned, this does not require the use of restriction enzymes or bisulfite reagent.

Several service providers on Genohub offer targeted bisulfite-seq, reduced representation bisulfite-seq (RRBS), methylated DNA immunoprecipitation seq (MeDIP) and whole genome bisulfite-seq (WGBS) library preparation and sequencing services. Simply click on one of these application types to get started. 

AGBT 2015 Summary of Day 3

Advances in Genome Biology and Technology Conference 2015

Day 3 of the Advances in Genome Biology and Technology meeting in Marco Island began with an announcement that next year the meeting would be held in Orlando due to hotel renovations, eliciting a groan from the audience. The meeting will come back to Marco Island in 2017.

Today’s plenary session speakers all presented work with a clinical focus, acknowledgement by the conference organizers about the direction of genome sequencing. The first speaker, Gail Jarvik, head of medical genetics at the University of Washington Medical Center presented on lessons learned from the Clinical Sequencing Exploratory Research (CSER) Consortium, marketed as ‘Hail CSER’. CSER is a national consortium of projects aimed at sharing innovations and best practices in the integration of genomic sequencing into clinical care. CSER has established a list of 112 actionable genes, some overlapping with the American College of Medical Genetics (ACMG) list. The CSER group annotated pathogenic and novel variants of the Exome Variant Server (EVS) to estimate rates in individuals of European and African ancestry.

The next talk was by Euan Ashley on moving toward clinical grade whole genome sequencing. He started by describing the genome as complex, full of repeats, duplications and paralogous sequences, giving him ‘a cold sweat at night’. He gave an example of a study with 12 adult participants who underwent WGS and described how clinical grade sequencing demands consistency in reporting. Most variants annotated as pathogenic were downgraded after manual review, but this takes lots of time. 12 individuals with 1,000 variants took around 1 hour per variant. In this case the use of WGS was associated with incomplete coverage of inherited disease genes, low reproducibility of detectable genetic variation and uncertainty about clinically reportable findings. He commented that new algorithms would be needed to address these problems and that ‘we’re at the beginning of genomics medicine’. Parts of his talk can be seen in his presentation at PMWC last month.

The last presentation before the break was by Levi Garraway who discussed the goal of cancer precision medicine to develop new therapeutics and combinations against molecular defined tumors. He mentioned that there are many discovery opportunities in clinical cancer genomics especially in terms of response and resistance to new therapies. Garraway sequenced the genomes of 57 prostate tumors and matched normal tissues to study somatic alterations. His model suggests that chormoplexy induces considerable genomic derangement over a relatively few number of events in prostate cancer supporting a model of punctuated cancer evolution. He introduced a 10X Genomics approach for phasing of large – 100 kb regions with exonic baits to obtain rearrangement information for chromoplexy. In the end he emphasized the importance of RNA-Seq profiling in conjunction with DNA sequencing for translational medicine to be relevant.

After the break, Stephen Kingsmore gave a presentation on rapid genome sequencing for genetic disease diagnostics in neonatal intensive care units. Kingsmore began the talk by describing how newborn screening (NBS) and early diagnosis reduces morbidity and mortality. NGS of 60 genetic diseases identifies ~5,000 affected newborns each year. He described how rapid genome sequencing (RGS) has the potential to improve NBS to most genetic diseases in newborns admitted to level II-IV NICUs. He mentioned a ‘ultra rapid’ sequencing pipeline he developed along with Illumina that takes 28 hours to go from sample to variant annotation (not publically available). He also discussed NSIGHT, a consortium for newborn sequencing sponsored by the NIH to understand the role of genome sequencing. More details can be found on the NHGRI page.

The last two plenary talks were by Christian Matranga and Malachi Griffith. Matranga described the clinical sequencing of viral genomes as important to understanding the evolution and transmission of the pathogen and the ability to inform on surveillance and therapeutic development. They developed a sequencing approach that combines RNAse H based depletion of rRNA with random primed cDNA RNA-seq to detect and assemble genomes from divergent lineages. They sequenced ~300 Lassa (LASV) and ~100 Ebola (EBOV) genomes. We describe some of their efforts in an earlier post called, Sequencing Suggests the Ebola Virus Genome is Changing. Be sure to read the New Yorker reference, it’s compelling!

Griffith’s talk was on optimizing genome sequencing and analysis. He makes the point that while most tumors are sequenced by exome sequencing at 75-100x mean coverage or by whole genome sequencing (WGS) to 30-50x mean coverage, detection of low frequency mutations require greater depth. He performed deep sequencing of an acute myeloid leukemia (AML) by WGS up to 350X, whole exome to 300X and using a capture panel of ~260 recurrently mutated AML genes to ~10,000x coverage. He found that deeper sequencing revealed more driver variants and improved the assignment of variants to clonal clusters. Checkout his animation of WGS depth down-sampling.

After lunch began the ‘Bronze sponsor workshops’, essentially the talks you pay >$40K to give. The most interesting was the last by 10X Genomics, mainly because as @bioinformer put it, “10X Genomics is the new princess of the AGBT ball”. First, check out the video that received a round of applause from the AGBT crowd: Changing the Definition of Sequencing. They announced their instrument would be available in Q2 this year, cost ~$75K and $500 / sample. This brings the question whether 10X Genomic’s microfluidic platform offers greater potential than Molecule. What are the implications for Illumina or PacBio? To learn more check out Keith Robison’s insightful post detailing all there is currently known about 10X Genomics.

After dinner began concurrent sessions on technology, genomic medicine and transcriptomics. Hopefully someone else will post details about the genomic medicine and transcriptomics sessions. The technology session began with Iain Macaulay describing G&T-seq, separation and parallel sequencing of genomes and transcriptomes of single cells. This was the first talk this year at AGBT with an embargo, no tweets were allowed. So rather than go into details, we did find this lecture online. The next talk was by Alexandre Melnikov on MITE-Seq, an approach to site directed mutagenesis referred to as Mutagenesis by Integrated TiLEs. MITE facilitates structure-function studies of proteins at higher resolution than typical site directed approaches. To read more check out their paper published last year in Nucleic Acid Research. Andrea Kohn, then described single-cell methylome profiling of Aplysia neurons. Using methyl-dip and bisulfite sequencing she achieved >20x coverage for each neuron and then added RNA-seq providing the first methylome and transcriptome from a single neuron. Next up was Sara Goodwin who gave an in depth analysis of the Oxford MinION Device for de novo and cDNA sequencing. She sequenced the yeast strain W303 to over 120x coverage and was able to achieve up to 80% aligned reads. She mentioned that identifying the right aligner was still a work in progress but overall found promise in the technology for long read sequencing, de novo assembly and splice site id.

Tomorrow’s plenary talks are the second installment of genomics, ‘Genomics II’ with presentations by Michael Fischback, Rob Knight, Chris Mason, and Gene Myers, excellent lineup to close the final day of AGBT. Checkout our earlier posts if you’ve missed day 1 or day 2

AGBT 2015 – Summary of Day 1

AGBT 2015

Highlights of Day 1 at AGBT

‘Welcome to paradise’, first words by Rick Wilson kicking off the annual Advances in Genome Biology and Technology (AGBT) meeting.  The plenary session began with David Goldstein from Columbia University presenting, “Toward Precision Medicine in Neurological Disease”.  David’s talk began with a discussion of clinical sequencing for neurological diseases, specifically large scale gene discoveries in epileptic encephalopathies. In epilepsy, 12% of patients are ‘genetically explained’ by a casual de novo mutation, which allows for the application of precision medicine. He discussed how a K+ channel plays a key role in at least two different epilepsies and how Quinidine which has never been used for epilepsy, was being used as a targeted treatment. He cautioned that in the literature there are too may correlation studies that don’t really amount to much and as we use genetics to target diseases, it’s critical to perform proper genetics driven precision medicine and not put patients on wrong treatment plans. To better characterize the effects of mutations, he emphasized the need for solid model systems. He also mentioned that he believes truly complex diseases can be tackled with enough patients, numbers matter. To illustrate his point, he described the sequencing of over 3,000 ALS patients to get a clear picture of what genes/proteins have therapeutic importance. At the end of his talk he was asked the old whole genome sequencing (WGS) vs. whole exome sequencing (WES) question and replied that WES was sufficient, as WGS added little due to lack of interpretability. This touched off some debate in the audience and Twitter with regards to Exome-seq and WGS. Highlighted are the advantages of each approach here: https://blog.genohub.com/whole-genome-sequencing-wgs-vs-whole-exome-sequencing-wes/.

The second talk in the plenary session was by Richard Lifton, from Yale and it was titled, “Genes, Genomes and the Future of Medicine”. Richard cautioned the audience, describing the rush to sequence whole genomes as more industry driven than good science, essentially reiterating the point that WGS is hard to interpret. This began a side discussion on Twitter about those who agree and disagree with this sentiment. Most notably, Gholson Lyon referenced two recent papers that demonstrated new ways to make processing of WGS data easier: and that the accuracy of INDEL detection was greater in WGS compared to WES, even in targeted regions. On the cost front, WGS at 35x coverage currently costs $1,750: https://genohub.com/shop-by-next-gen-sequencing-project/#query=ef95a222ca23fc310eedf6de661e4b22 or $3,500 for 70X coverage, while whole exome sequencing costs at 100x coverage are around $1,314: https://genohub.com/shop-by-next-gen-sequencing-project/#query=0d4231a1d12425085f4e284373605acd.  Richard remarked at the end of his talk that not much had been found in non-coding regions, several in the audience challenged him on this assessment.

The third talk in the plenary session was by Yaniv Erlich titled, “Dissecting the Genetic Architecture of Longevity Using Massive-Scale Crowd Sourced Genealogy”. We’ve had the pleasure to hear Yaniv give several lectures in the past, all have been engaging, this was no different. His talk was on using social media to dissect the genetic architecture of complex traits, specifically whether they work independently (additive) or together (epistatic). Predictions of epistasis suggest an exponential increase with added risk alleles. He used geni.com to dissect complex traits in large family trees and validated publicly submitted trees using genetic markers. He encoded birthplace as GPS coordinates using Yahoo Maps and showed migration from the Middle Ages through the early 20th Century. The video he played was amazing, check it out: https://www.youtube.com/watch?v=fNY_oZaH3Yo#t=19. His take home message was that longevity is an additive trait, which is good for personalized medicine. The Geni data he described is open to the public for use.

The fourth and final talk of the night was by Steven McCarroll titled, “A Common Pre-Malignant State, Detectable by Sequencing Blood DNA”. He started by posing the questions, what happens in the years before a disease becomes apparent; cancer genomes are usually studied when there are enough mutations to drive malignancy, do they happen in a particular order? He examined 12,000 exomes for somatic variants at low allele frequency and uncovered 3,111 mutations. Blood derived schizophrenia clustered in four genes: DNMT3A, TET2, ASXL1 and PPM1D, all disruptive. He postulates that driver mutations give cells an advantage, over several years clonal progeny takes over creating pre-cancerous cells. Therefore clonal hematopoiesis with somatic mutations can be readily detected by DNA sequencing and should become more common as we age. Patients with clonal mutations have a 12 fold higher rate of blood cancer, meaning there is a window for early detection, possibly 3 years. This work was recently published in the New England Journal of Medicine: Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence.

Today’s sessions were impressive and set expectations high for tomorrow’s talks. McCarroll’s last comment nicely captured this sentiment, setting the tone for the rest of the meeting, “Medicine thinks of health and illness, there is a lot in between that is ascertainable via genome sequencing”. 

Whole Genome Sequencing (WGS) vs. Whole Exome Sequencing (WES)

Gene, exon, intron, sequencing

 

“Should I choose whole genome sequencing (WGS) or whole exome sequencing (WES) for my project?” is such a frequently posed question during consultation on Genohub, we thought it would be useful to address it here. With unlimited resources and time, WGS is a clear winner as it allows you to interrogate single-nucleotide variants (SNVs), indels, structural variants (SVs) and copy number variants (CNVs) in both the ~1% part of the genome that encodes protein sequences and the ~99% of remaining non-coding sequences. WES still costs a lot less than WGS, allowing researchers to increase sample number, an important factor for large population studies. WES does however have its limitations. Below we’ve highlighted the advantages of WGS vs. WES and described a real case example of someone ordering these services using Genohub.

Advantages of Whole Genome Sequencing

  1. Allows examination of SNVs, indels, SV and CNVs in coding and non-coding regions of the genome. WES omits regulatory regions such as promoters and enhancers.
  2. WGS has more reliable sequence coverage. Differences in the hybridization efficiency of WES capture probes can result in regions of the genome with little or no coverage.
  3. Coverage uniformity with WGS is superior to WES. Regions of the genome with low sequence complexity restrict the ability to design useful WES capture baits, resulting in off target capture effects.
  4. PCR amplification isn’t required during library preparation reducing the potential of GC bias. WES frequently requires PCR amplification as the bulk input amount needed to capture is generally ~1 ug of DNA.
  5. Sequencing read length isn’t a limitation with WGS. Most target probes for exome-seq are designed to be less than 120 nt long, making it meaningless to sequence using a greater read length.
  6. A lower average read depth is required to achieve the same breath of coverage as WES.
  7. WGS doesn’t suffer from reference bias. WES capture probes tend to preferentially enrich reference alleles at heterozygous sites producing false negative SNV calls.
  8. WGS is more universal. If you’re sequencing a species other than human your choices for exome sequencing are pretty limited.

Advantages of Whole Exome Sequencing

  1. WES is targeted to protein coding regions, so reads represent less than 2% of the genome. This reduces the cost to sequence a targeted region at a high depth and reduces storage and analysis costs.
  2. Reduced costs make it feasible to increase the number of samples to be sequenced, enabling large population based comparisons.

Most functional related disease variants can be detected at a depth of between 100-120x (1) which definitely makes the cost case for exome sequencing. Today on Genohub if you want to perform whole human genome sequencing at a depth of ~35X, the cost is roughly $1700/sample. If you were to request human exome-sequencing services with 100x coverage, using a 62 Mb target region, your cost would be $550/sample. Both of these prices include library preparation. So in terms of producing data WES is still significantly cheaper than WGS. It’s important to note that this doesn’t include your data storage and analysis costs which can also be quite a bit higher with whole genome sequencing.

It’s also important to remember that depth isn’t everything. The better your uniformity of reads and breath of coverage, the higher the likelihood you’ll actually find de novo mutations and call them. And that’s the main goal, if you can’t call SNPs or INDELs with high sensitivity and accuracy, then the most high depth sequencing runs are worthless.

To conclude, whole genome sequencing typically offers better uniformity and balanced allele ratio calls. While greater exome-seq depth can match this, sufficient mapped depth or variant detection in specific regions may never reach the quality of WGS due to probe design failures or protocol shortcomings. These are important considerations when examining tissues like primary tumors where copy number changes and heterogeneity are confounding factors.

If you’re ready to start an exome project, spend a few minutes determining the coverage you’ll need for your experiment. We have an exome-seq guide with examples to help you determine the number of sequencing reads you need to achieve a certain coverage of your exome. If you’re planning to embark on whole genome sequencing, use our NGS Matching Engine which automatically calculates the amount of sequencing capacity on various platforms to meet the coverage requirements for your project.

Reference:

1) Effect of Next-Generation Exome Sequencing Depth for Discovery of Diagnostic Variants

Illumina’s Latest Release: HiSeq 3000, 4000, NextSeq 550 and HiSeq X5

HiSeq 3000, HiSeq 4000, HiSeq X Five, HiSeq X Ten

Illumina’s latest instrument release essentially comes down to more data/day.  Using the same patterned flow cell technology already in use with the HiSeq X Ten, The HiSeq 3000 has an output of 750 Gb or 2.5B PE150 reads in 3.5 days. The HiSeq 4000 has two flow cells, so twice the output: 1.5 Tb, 5B PE150 reads in 3.5 days. The NextSeq 550 combines the current NextSeq 500 with a microarray scanning system that fits right into the flow cell holder. The HiSeq X Five is less exciting; just half the number of instruments as the HiSeq X Ten.

If you don’t have the $10M budget for a HiSeq X Ten, you can purchase a HiSeq X Five and scale to the X Ten at a lower price/instrument: $1M/unit

Price Price/unit $/Genome* Consumables $/Gb
HiSeq X Five $6M $1.2M $1,425 $1,200 $10.6
HiSeq X Ten $10M $1M $1,000 $800 $7

*Price per 30X human genome according to Illumina. We’re not aware of any sequencing facility currently offering 30 human genomes for $1,000. On Genohub today, you can order a single whole human genome at 35X for $1,750.

Both the HiSeq X Five and Ten are still only “licensed” for human whole genomes [Update: Since this post was published in January 2015, Illumina now allows the sequencing of other large species on the HiSeq X Ten. For an up to date status on what is and what isn’t allowed on a HiSeq X, follow our HiSeq X Guide Page]. That basically means that while they can technically be used on non-human samples or transcriptomes, Illumina wants these focused on the WGS market (probably thinking about the BGI / Complete Genomic’s WGS instrument release this year).  Plus it gives them an excuse to release patterned flowed cells on more models, hence the HiSeq 3000/4000. Interestingly, Illumina is going to start bundling the TruSeq PCR-free and TruSeq Nano library prep kits (the only chemistry currently compatible with the X Five and X Ten) with X Five/Ten cluster reagents. At least for now, they don’t intend on doing this with the HiSeq series. Other news from this release:

HiSeq 3000/4000 do not have a rapid mode, high-output only. However PE150 reads only take 3.5 days

You can’t upgrade from a HiSeq 2500 (non-patterned flow cell) to a HiSeq 3000 or 4000 (patterned flow cells)

You can upgrade from the single flow cell HiSeq 3000 to dual flow cell HiSeq 4000

HiSeq 3000 yields >200 Gb/day, a 28% increase vs. HiSeq 2500 v4, yet the cost to purchase a HiSeq 3000 is the same as a HiSeq 2500. With two flow cells, the HiSeq 4000 yields twice as much data.

Sequencing Applications and Turnaround Time 

Exomes Transcriptomes 30X Genomes
HiSeq 3000 90 (2×75, <2 days) 50 (2×75, <2 days) 6 (2×150, 3.5 days)
HiSeq 4000 180 (2×75, <2 days) 100 (2×75, <2 days) 12 (2×150, 3.5 days)

So in the end, assuming sequencing facilities aren’t fed up with this break neck upgrade cycle and actually purchase these instruments, researchers can expect more data with faster turnaround times. We’ve already spoken to a few of our service providers who are considering upgrades to their HiSeq 1500/2000 instruments. As soon as these new instruments are available on Genohub, we’ll make an announcement [Update: they are all available, use our NGS Matching Engine for access to the latest Illumina instruments]. If you’d like to be the first to know send us an email at support@genohub.com. In the meantime, our providers offer services on the HiSeq 2500 v4, HiSeq X Ten, NextSeq 500 and HiSeq instruments (amongst many others).  You can order these services immediately and expect data delivery within the listed guaranteed turn around times. If you’re not sure what technology / instrument is right for you, just enter the number of reads or coverage you need and let our NGS Matching Engine identify the best service for you.  So what’s next? A little bird has told us patterned flow cells on the MiSeq !