[WhatsOEver]

From what I have seen they can generate a ton of raw data and some sequence (bacterial genomes), but the quality per read is not impressive (yet). It relies on a polymerase which is much slower than the helicase ONT uses so yield will be lower unless they can have a much higher number of pores, and while polymerases can be highly accurate they also have a much larger variation in incorporation time so signal processing seems very challenging.

How does this fit together?
Could you tell us where you got this information from?
I would especially be interested in seeing some real data from this machine.

[gringer]

Is this human data? Could you share the raw data behind this? This might be an interesting region to evaluate our aims but I would need to check whether we are able to target this region with our approach.

This is parasite DNA, but structures like this exist in human chromosomes as well, particularly in the centromeric regions. The FASTQ files associated with this (called prior to the homopolymer fixes) can be found here:

https://www.ncbi.nlm.nih.gov//sra/?term=SRP092357

I'm still working on getting the raw [signal-level] data uploaded so that people can use their own calling algorithms. SRA can't handle nanopore raw signal, so I'll need to switch to EBI/ENA for that.

I might have a look at the NA12878 genome (or Clive Brown's genome) later on to see if anything similar can be found there (adding to my list of things to do once I get back home).

[WhatsOEver]

I might have a look at the NA12878 genome (or Clive Brown's genome) later on to see if anything similar can be found there (adding to my list of things to do once I get back home).

That would be great! Thanks! I'll have a look at the RepliG generated worm data in the meantime.

[ymc]

Just curious. Is it possible to use Minion to sequence a single cell without any amplification at all?

[WhatsOEver]

Just curious. Is it possible to use Minion to sequence a single cell without any amplification at all?

According to the information we got from the ONT developers it is not possible. And even if it would be, you will end up with 1X (or 2X for diploid...) only with no possibility to do error correction (consensus building). You will also not be able to do any repetition - if not all DNA is sequenced (for what ever reason) it is lost. That being said, we will for sure try it sooner or later if decide to buy one

[Brian Bushnell]

Why do you think so for human (or more generally multiploid organisms) WGS? For human data, we are currently unable to do whole genome reconstruction with short reads alone using the existing reference. If we create a scaffold of our genome of interest with long reads, we would still be unable to map the short reads accurately. As an example: How would a dual platform approach help me to resolve highly repetitive regions in the genome like MHC or mucins?

The combination of library types would get you highly accurate SNP and short indel calls in the nonrepetitive majority of the genome from the short reads, with minimal coverage, and SV calls across the full genome. It's also possible to combine short and long reads to accurately phase the variants called by the short reads, and potentially call variants in repetitive areas using various techniques like correcting the long reads with the short reads.

[gringer]

you will end up with 1X (or 2X for diploid...) only with no possibility to do error correction (consensus building).

2X (or 4X for diploid); DNA is double-stranded, so there is a possibility of consensus sequencing using 1D².

[ymc]

According to the information we got from the ONT developers it is not possible. And even if it would be, you will end up with 1X (or 2X for diploid...) only with no possibility to do error correction (consensus building). You will also not be able to do any repetition - if not all DNA is sequenced (for what ever reason) it is lost. That being said, we will for sure try it sooner or later if decide to buy one

I am interested in working on samples with around 10 human cells that share the same DNA (barring mosaicism). It will be great if you also give this scenario a try and tell us if we can get any sequences from the ten cells.

[WhatsOEver]

@Brian Agreed on everything except

potentially call variants in repetitive areas using various techniques like correcting the long reads with the short reads.

You cannot map short reads containing only repetitive sequence to a long read spanning the repetitive region, can you? If you could, you would be able to map the same short read to the reference assembly - which you can't. Or am I missing something important in your statement?

@ymc
I would start with fresh cells from the cell line for testing something that risky... but sure, I'll get back to you after we have made progression

[gringer]

mapping short reads is easier once the complete assembly exists. You can at least use probabalistic methods for mapping, and distribute reads across all the locations they could potentially map to.

Bowtie2 does this by choosing one of the candidate locations at random for each read (as long as they are equivalently incorrect, and represent the best match). It ends up being the same thing as long as there is sufficient read depth and no systematic error in the reference assembly or short reads.

[Ola]

How does this fit together?
Could you tell us where you got this information from?
I would especially be interested in seeing some real data from this machine.

If you make 10000 measurements per second and read DNA at a slow speed it will generat lots of raw data but not so much sequence. They presented early data at a VIB conference. I would not make too much of it at this point, they showed a bit about how it works and that it can generate data. I don't think they aim for WGS, more like gene panels and specific clinical tests.

[gringer]

I am interested in working on samples with around 10 human cells that share the same DNA (barring mosaicism). It will be great if you also give this scenario a try and tell us if we can get any sequences from the ten cells.

Amplification will be needed from 10 cells as well. The lower limit currently for nanopore (allowing for a bit of a drop in yield) is about 10,000 cells.

The rapid sequencing kit needs about 200ng of high-quality DNA for sequencing reads of about 1-10kb, but it's possible to go down to about 10% of that before the drop in run yield falls off a cliff.

[Brian Bushnell]

You cannot map short reads containing only repetitive sequence to a long read spanning the repetitive region, can you? If you could, you would be able to map the same short read to the reference assembly - which you can't. Or am I missing something important in your statement?

I know it sounds odd, but yes, you can. Imagine two long reads covering different repeats (or different ploidies). Short reads cannot map to the assembly and differentiate which one they should map to, because in the assembly, they look the same. But if there is an actual SNP in one copy, then reads with that SNP will (in general) map to the copy with the SNP, and reads without the SNP will (in general) map to the copy without the SNP. It's not perfectly clean due to the noisiness of the long reads, but with sufficient copies, on average, the version with the SNP will get corrected to have the SNP, and the version without the SNP will get corrected to the reference. Then, aligning the corrected long reads will allow you to properly place the SNP in the correct copy. I've done this with Poplar PacBio + Illumina libraries for the purpose of phasing.

[gringer]

Here's one 222kb read [6a6944fa-9143-490f-b5dd-22e7e7267d3c_Basecall_1D_template] from the nanopore-wgs sequencing project, run FAF09968. The resolution of this is 109 bases per pixel, so the repeat unit length is around 2500bp:

http://i.imgur.com/5MNzBsf.png

Smaller version:



Unfortunately, I haven't yet been able to find a completely covered repeat, but hopefully this is sufficient to demonstrate the sequencing issue with other platforms.

For additional reference, here is the WGS consortium github page:

https://github.com/nanopore-wgs-consortium/NA12878

Direct link to this run fastq file:

http://s3.amazonaws.com/nanopore-human-wgs/rel4-nanopore-wgs-3439856925-FAF09968.fastq.gz

The read itself can be found here:

http://www.gringene.org/data/largeRepeat_6a69.fa.gz

[gringer]

Unfortunately, I haven't yet been able to find a completely covered repeat

... and I just found one (308kb read). The main repeat region has unique sequence outside it (e.g. where that line of blue is):

http://i.imgur.com/aym9zTg.png

Read is here:

http://www.gringene.org/data/largeRepeat_6cec.fa.gz