FDA approves 'next gen' DNA sequencers

[dasmoocher]

@Smeghead

Secondly, gene expression is certainly a very different test from genomic sequencing. Genomic information looks at the sequence of the cells’ DNA. Gene expression testing measures relative levels of specific mRNAs. Extracting RNA from a sample and counting it is a very different process from extracting DNA and reading it. Expression testing lets you see, for instance, that sample X has twice as many CFTR mRNAs per cell than sample Y. That’s informative for some diseases, like cancer, but not really for something like CF.

Be careful, that type of explanation can be condescending and biased.

[deltasigma]

@dasmoocher

Whatever. Is your next act going to be telling a statistician what the word “significant” means?

Did you miss my admission of grievous an irredeemable wrongdoing? Here, let me play it again for you in slow motion, so you can really savor it.

@deltasigma

That was a poor choice of words in post 11 - I admit that.

[dasmoocher]

I’m done with that sidetrack.

I think a more interesting question with regard to the topic of the OP is which company will get their platform/assays FDA-approved next?

Illumina is the NGS market leader, but other companies have different approaches and technologies.

[deltasigma]

Here is the list wikipediahas. I’m sure you will be happy to critique.

[lazybratsche]

@dasmoocher

I believe you can target specific regions when the cDNA library that will be sequenced is made. You make the library from your source DNA–adding sequences that can act as PCR primers and barcodes to sheared fragments of the preferential size, basically, and it’s this library DNA that is sequenced.

Using their Nextera Rapid Capture Custom Enrichment Kit, they say you can:
So, you don’t have to sequence the whole genome, but instead you can look at regions of interest that you “enriched” in making your library:

So, for this FDA-approved assay, it looks like you sequence the region containing the CFTR gene.

This make sense for a assay looking for variants in a particular gene. Instead of wasting reagents on the rest of the genome, you sequence the same region many, many times. Errors are made in making the libraries–noise, due to the error rate of the enzymes, and with deeper coverage you can get a stronger signal to address whether that SNP is a mis-incorporation of an nucleotide by the polymerase, or a real variant that shows up multiple times in the exact same spot.

I haven’t done NGS myself, but I image early errors would be more noisy, though.

Gotcha. That makes sense, there are certainly no shortage of other pull-down techniques to enrich for DNA of interest. And I see that the CFTR locus is over 100 kb in size which explains why Sanger sequencing isn’t a great choice in this case.

[dasmoocher]

@Smeghead

You could, also, look at DNA samples for epigenetic marks, which would tell you something about how that gene is being expressed, but I don’t think our understanding of those systems are good enough for this to really tell us anything clinically important at this point.

There are some methylation imprinting disorders where this could be useful, such as Beckwith Wiedemann syndrome, and Angelman and Prader-Willi syndromes.

With Illumina’s MiSeq plaform, since you’re sequencing a library made with un-modified bases, you’d have to make and sequence a bisulfite-converted library to see where the methylated cytosines are compared to a non-treated library. I hear this can be labor intensive on the bench.

Pacific Biosciences and Oxford Nanopore Technologies, for example, have platforms that are supposed to be able to determine base modifications directly.

PacBio uses the differences in base incorporation kinetics to do this (Base Modification Detection). It looks like it takes longer to incorporate the matching nucleotide to a modified base than an unmodified base and this shows up in the fluorescent trace.

Oxford Nanopore, and I don’t think people have gotten a great look at their data since it’s very new on the market, has a platform where the DNA is threaded through a nanopore in a membrane and the electrical conductance through the nanopore changes with what nucleotide is in the pore.

Nat Nanotechnol. 2009 Apr;4(4):265-70. doi: 10.1038/nnano.2009.12. Epub 2009 Feb 22.
Continuous base identification for single-molecule nanopore DNA sequencing.

Abstract

A single-molecule method for sequencing DNA that does not require fluorescent labelling could reduce costs and increase sequencing speeds. An exonuclease enzyme might be used to cleave individual nucleotide molecules from the DNA, and when coupled to an appropriate detection system, these nucleotides could be identified in the correct order. Here, we show that a protein nanopore with a covalently attached adapter molecule can continuously identify unlabelled nucleoside 5’-monophosphate molecules with accuracies averaging 99.8%. Methylated cytosine can also be distinguished from the four standard DNA bases: guanine, adenine, thymine and cytosine. The operating conditions are compatible with the exonuclease, and the kinetic data show that the nucleotides have a high probability of translocation through the nanopore and, therefore, of not being registered twice. This highly accurate tool is suitable for integration into a system for sequencing nucleic acids and for analysing epigenetic modifications.

My bold

These platforms are probably aren’t close to FDA approval for clinical tests, though. And there’s a lot more to epigenetics than modified DNA bases.

[Smeghead]

@dasmoocher

There are some methylation imprinting disorders where this could be useful, such as Beckwith Wiedemann syndrome, and Angelman and Prader-Willi syndromes.

There are already methylation-specific PCR tests on the market for detecting Angelman and Prader-Willi syndromes.

I know because I used to do them.

There’s no need to step up to next-generation sequencing for these, because you only need to check a relatively short stretch of DNA for methylation. Next-gen would be needless complication.

[dasmoocher]

@Smeghead

There are already methylation-specific PCR tests on the market for detecting Angelman and Prader-Willi syndromes.

I know because I used to do them.

There’s no need to step up to next-generation sequencing for these, because you only need to check a relatively short stretch of DNA for methylation. Next-gen would be needless complication.

Good point, if there’s an easy test to for specific targets. I was just thinking off the top of my head of how you could apply NGS for detecting methylation status, so maybe Angelman and Prader-Willi aren’t the best future clinical test examples, if cheaper and easier tests exist.

Out of curiosity, what are the PCR-based assays you used? Were these using specific primers after bisulfite conversion? I know Qiagen has PCR assays for CpG islands based on restriction enzyme digests. I haven’t checked them out lately, but I found their “% of methylation” calculation to be murky. Is that “%” of methylated DNA templates (50% of the template molecules are methylated vs unmethylated), or “%” of methylation within a template molecule CpG island, or both and how can you tell by looking at Ct values. I called their tech support once years ago when they first came out (SABiosciences, actually; later acquired by Qiagen) and didn’t really get a satisfactory answer, but this might be clearer now since last time I asked.

What’s nice about the PacBio and Oxford Nanopore technology is you take bisulfite or methylation-specific digests out of it. But these aren’t FDA-approved platforms yet.

I think the real utility of NGS, though, is that–unlike making PCR primers or hybridization probes for a microarray for detection of variants, you don’t need to know the sequence beforehand.

What NGS could be useful for is genome wide association studies for SNPs that contribute to multifactorial diseases.

Figure 1 in this article (Chapter 11: Genome-Wide Association Studies) is a graph of Effect Size (Odds Ratio) versus Allele Frequency.

Interesting that our FDA-approved CFTR gene is used as their example of rare(r) variants having a very large effect in a specific disease. NGS can be useful for identifying common variants that have much less effect individually but are associated with something more general like cardiovascular disease that’s not a classic, easy to follow example of Mendelian inheritance.

It is important to note that the technology for measuring genomic variation is changing rapidly. Chip-based genotyping platforms such as those briefly mentioned above [Illumina and Affymetrix microarrays] will likely be replaced over the next few years with inexpensive new technologies for sequencing the entire genome. These next-generation sequencing methods will provide all the DNA sequence variation in the genome. It is time now to retool for this new onslaught of data.