Name: Hans Scheffer
Position: Associate professor of clinical molecular genetics (since 2009) and head of the Division of DNA Diagnostics (since 2001), Department of Human Genetics, Radboud University Nijmegen Medical Centre, the Netherlands
Experience and Education:
Assistant professor, Department of Medical Genetics, University of Groningen, 1988-2001
PhD in molecular genetics, University of Groningen, 1990
MSc in pharmacy, University of Groningen, 1983
Hans Scheffer coordinates a European project called Techgene that aims to use massively parallel sequencing technologies to develop, optimize, and implement diagnostic tools for genetic disorders.
The three-year project, which is funded with €3 million under the European Union's 7th Framework Program and officially started in February, involves 12 groups from nine European countries and is in the midst of testing several second-generation sequencing platforms for diagnosing genetic diseases.
In Sequence spoke last week with Scheffer, who heads the Division of DNA Diagnostics in the Department of Human Genetics at Radboud University Nijmegen Medical Centre. Below is an edited transcript of the conversation.
Tell me about Techgene. What is the aim of this project?
The main focus of Techgene is to get next-generation sequencing implemented and validated in human genetic diagnostics. We are a consortium of 12 laboratories and commercial enterprises.
For a couple of selected model genetic disorders with increasing complexity, we have designed some issues that we want to address. We plan to use a variety of enrichment strategies, including amplicon sequencing, array-based enrichment, and liquid enrichment, so for a selected number of disorders, we can pull out the relevant genes from the complete genome and then sequence those genes in parallel.
The simplest genetic disorder we have selected is hereditary breast cancer, where there are two genes involved that we will address [BRCA1 and BRCA2]. Also, the hemoglobinopathies are included in this category.
A more complex group of disorders are the sensory disorders — hereditary blindness and hereditary deafness — and also Usher syndrome. Our rough estimate is that, for instance, for hereditary blindness, there are something like 150 genes involved. A comparable complexity is envisaged for movement disorders, the paraplegias and ataxias.
The last category that's very complex is mental retardation. In that category, copy number variants are also involved; it's not just sequence variants. That's really rather complex for us as diagnostic laboratories right now.
We are also involved in the validation of the developed approaches. That's done in connection with another European project, a Network of Excellence called EuroGentest. Some of our partners are involved in EuroGentest as well. It's a network of more laboratories that are involved in increasing the quality in clinical genetic laboratories and clinical genetics in general.
The final issues we would like to address are economic issues — what does it cost? There is also a work package on ethical issues, for instance, how to deal with genetic information that's not requested.
Who are the participants?
It's a mixture of renowned hospital laboratories or reference laboratories and also research laboratories that are well known for their front-running activities in next-generation sequencing.
[ pagebreak ]
Can you tell me about your own laboratory, the Division of DNA Diagnostics at Radboud University Nijmegen Medical Centre?
We consider ourselves as one of the frontrunners in diagnostic development. One parameter is the huge number of genes that we test, at least for an academic laboratory. We offer [testing for] something like 250 different genes.
How did you decide to explore the new sequencing technologies? What can they bring to the table that Sanger sequencing cannot?
What we now see in clinical genetic testing for humans that have a hereditary disorder is that what we are quite good at as laboratories in general is a hereditary disorder in which there is a correlation of one disease, one gene. For instance cystic fibrosis, that can be done quite efficiently.
What we cannot do efficiently is genetic disorders in which we cannot see from the phenotype, or from the clinical characteristics a patient has, which gene is involved. The example could be hereditary blindness — you cannot see from the outside which gene is involved. Although we know that maybe up to 150 genes could cause the disorder, we are now sometimes in the situation that we are requested to sequence one gene, and then, subsequently, if the result is negative, we get a request for sequencing a second gene. That's not a good situation for a patient.
This is the next step. We can do what we call "genetically heterogeneous disorders." We can now offer a diagnostic approach in which we analyze [several] genes in parallel.
And, though this is not the aim of this project, this is going to [be applied to] common disorders, in which it is known that there is a substantial genetic contribution to the etiology — for instance, diabetes or Parkinson's or Alzheimer's. In theory, in the near future, if we can do a risk analysis of those factors, this technique has the power to also address those disorders. That's not the aim of this project, [but] definitely a step towards that approach.
Which sequencing platforms are you testing as part of Techgene?
All platforms are available in Techgene. In the proposal, the 454 Titanium platform, the Illumina platform, and the [Applied Biosystems] SOLiD are mentioned. The SOLiD and the 454 Titanium are available in my laboratory, and the Illumina is available in a couple of partner laboratories. The Helicos is not mentioned in the proposal itself, but it is in one of the groups [at Leiden University], so it's available in the consortium.
Obviously, we are comparing the platforms for a number of parameters: throughput, turnaround time, quality of the sequences that come out, the ease with which sequence data can be interpreted, costs.
Are any of the vendors of sequencing platforms involved in the consortium?
We have what we call "affiliated partners" — partners that contribute to the discussions in the knowledge network [but don't contribute to the budget]. We have 454 and Agilent involved in those; we collaborate closely with them. Also, we have close connections with RainDance and with Nimblegen.
What are the challenges the new sequencing platforms, in general, pose for their use in diagnostics?
One of the challenges is to have sufficient coverage, for instance, to be able to identify heterozygous mutations, or to identify mosaicism, or copy number variants. So to quantify, that's a challenge.
Another one is to develop a pipeline for the interpretation of results. That will definitely be one of the major bottlenecks that we envisage. For those issues, we have established collaborations with companies that are developing tools for data analysis. CLC Bio is one of the companies.
[ pagebreak ]
What about other potential challenges, such as turnaround time and sample preparation, which some say is cumbersome?
The turnaround time — the time it takes a laboratory to do all the preparations — that's definitely an issue that has to be improved. [However], I cannot confirm that the procedures are cumbersome. We have the enrichment protocols [for the NimbleGen approach] now in house, and I guess I can say that we can perform them in a robust way for both platforms. We are also now getting — that's in a more preliminary stage — experience with liquid hybridization with the RainDance approach.
But you are definitely right, for diagnostic purposes, it takes too long. The run time itself, that's no problem for the 454 platform, but it's definitely a drawback, for instance, for the SOLiD platform and maybe also for the Illumina platform. However, on the other hand, the relative costs are much higher for the 454 platform.
What kinds of improvements do you expect to happen over the next year or so?
The costs will be definitely reduced, and the throughput will be increased substantially.
Also, [we will] be able to perform the analysis in a multiplexed way, to analyze multiple samples in parallel. That's another issue we try to address in Techgene — the possibility to barcode the samples and analyze them in parallel, and then afterwards, being able to identify, due to the barcodes, which results come from which sample. This also makes the overall procedure much more efficient. That's a novelty in diagnostics — we are quite used to work in a way that one sample relates to one patient, and now we run samples in one experiment in parallel. That puts some constraints on the quality issues [but] we are quite confident that that can be resolved.
Are you already using your in-house next-gen sequencers for routine diagnostic applications?
First, we want to do the development and the validation processes quite separately, so that we know for sure that the results that are coming out are definitely what we believe them to be. We are currently in the development stage and are entering the validation stage for a couple of applications right now. But [next-gen sequencing is] not offered for diagnostic purposes right now. That will be the case for a couple of applications within four to six months — then we will definitely be able to offer the first diagnostic applications.
What will these applications be, and what are the advantages of switching them over to next-gen sequencing?
It will definitely be for the more simple disorders. The main advantage [compared to Sanger sequencing] would be that we analyze more sequence per patient, so the sensitivity is increased. And it saves time because we can analyze the genes in parallel.
Obviously, Sanger sequencing can be done quite efficiently, but we are put in a situation that we have to run genes sequentially instead of in parallel. The run time of a next-generation sequencing run is much longer than a run on an automated Sanger sequencing machine, but there is the possibility to run genes in parallel instead of sequentially.
Considering the costs, that's something we have to study. We are not quite sure whether the cost will, at the moment, already be reduced compared to Sanger sequencing. That's definitely not far away, due to strategies like barcoding and running samples in parallel.
Are you going to consider new sequencing platforms as they become available?
Oh yes, definitely. We have an open eye on the current developments, which may offer additional advantages. We will try to include those platforms as well, although it's not in the proposal. But this group is moving in such a rapid way that it wouldn't make sense to just stick to what we have initially proposed. We have an open eye for all developments that might be useful for diagnostic applications.
How long until high-throughput sequencers will become a routine platform in genetic testing labs?
I guess it's not far away. It will be between 6 and 12 months, and then they [will be] available in at least a couple of reference laboratories, and then the rest will follow quite soon.
Do you think one platform will dominate, or will these labs adopt several platforms?
I guess there will be niches for the different platforms. That has to do with the different features, of course. For example, the Roche/454 platform has the long reads, that has an advantage in determining phase for different DNA variants you identify. And the SOLiD and the Illumina platforms have the advantage of deeper coverage, so those might be more suitable for mosaicism determination or copy number variant identification. And for sure, the Helicos platform will have another niche. There, the advantage is that amplification is not necessary.