Tagging long amplicons

[Sidsel]

Dear Seqanswers,

I'm very happy to join the community - and I'm very excited to see if anyone has some input to this first post of mine

I'm very new with NGS, and I'm starting out on the Illumina platform: miseq. I'm trying to set up an assay, but I simply can't figure out how to combine a protocol with:

1. really long amplicons (5-10 kb) which will be fragmented on the miseq platform
2. tagging of my fragmented amplicons to be able to sort them according to origin afterwards!

The deal is, we would like to look at entire genes and flanking regions from samples that have origin in different countries - we would like to cover approximately 37.500 bp per sample, and we have 1000 samples.

I would like to be able to tell apart which reads are from
1. which genes?
2. which country?

That's 3-400 samples from 3 different countries = 3 tags
PCR on 5 genes from each sample = another 5 tags

I'm doing long amplicons that need to be fragmented, because doing 400 bp amplicons would be almost 100 PCRs per sample...(1000 samples!)

I'm aware that I can generate tagged amplicons, by for example dual indexing - but when I do that, they'll be tagged in the 3' and 5' end, which leaves me with a lot of 250 bp reads with no tags, since the amplicons are 5-10 kb and will be fragmented, right??

I was hoping that there was some kind of kit that would allow me to tag the fragments when tagging them with the illumina adapters, but I can't seem to find that?

If anyone has any input - please post!!! I would be greatful for ANY feedback.
Even if you think miseq is simply the wrong platform!

Hugs
Sidsel

[mastal]

Hi Sidsel,

MiSeq sounds like the right platform, because you don't need too many reads from each sample.

I have been working with long amplicons, but only with the bioinformatics side of things, so I am not familiar with the Illumina protocols for processing amplicons.

I would imagine that what you can do is combine all the different amplicons (genes) from the same sample into one sample to be sequenced, then they would all (all the genes and all the reads) have the same Illumina dual-index barcode, I assume that you can get up to 96 or so different barcodes on one lane of the MiSeq.

The problem is that with the number of samples you have, that still works out quite expensive.

Hope this helps,
Maria

[Sidsel]

Dear Maria,

Thank you very much for your reply!

I apologize if these questions are almost too basic to even understand - but I'm going to try an ask them anyway

Since you are working with the bioinformatics, when you say that you've been working with long amplicons, does that mean that the reads that you received from the sequencer were long?? Or did you assemble the reads into long sequences?

I've understood that the miseq "only" generated reads of up to 250 bp? So long amplicons have to be fragmented and then assembled bioinformatically?

I guess I misunderstood the dual-indexing system, because I understood that the barcode is generated when doing the PCR on the genetic region BEFORE doing the actualy miseq procedure. This is why I thought that the long amplicons would only have the barcodes in the ends, and therefore generate alot of reads with no barcodes when fragmented prior to sequencing.

If the dual indexing system barcodes the amplicon fragments and not the actual intact amplicon, then I could perhaps just pool all the amplicons that are from the same gene and the same country - that would result in 30 pools, with amplicons from 100 samples in each.

Is that feasible? Would I be able to bioinformatically tell them apart afterwards? And would I be able to run that on one miseq run?

Thank you so much!
Sidsel

[mastal]

Since you are working with the bioinformatics, when you say that you've been working with long amplicons, does that mean that the reads that you received from the sequencer were long?? Or did you assemble the reads into long sequences?

No. I meant that our initial samples that we sent for sequencing were a bunch of long PCR amplicons, ranging in size from about 3.5 kb to perhaps 8 kb.

I've understood that the miseq "only" generated reads of up to 250 bp? So long amplicons have to be fragmented and then assembled bioinformatically?

Yes, that's right.

I guess I misunderstood the dual-indexing system, because I understood that the barcode is generated when doing the PCR on the genetic region BEFORE doing the actualy miseq procedure. This is why I thought that the long amplicons would only have the barcodes in the ends, and therefore generate alot of reads with no barcodes when fragmented prior to sequencing.

This is probably the way it works for Illumina's custom amplicon kits, which generate amplicons of the order of 400 bp, as you mentioned previously.
For the long amplicons that need to be fragmented, they would probably be treated as 'normal' DNA samples, so the Illumina barcodes would be added after the fragmentation step.

If the dual indexing system barcodes the amplicon fragments and not the actual intact amplicon, then I could perhaps just pool all the amplicons that are from the same gene and the same country - that would result in 30 pools, with amplicons from 100 samples in each.

Is that feasible? Would I be able to bioinformatically tell them apart afterwards? And would I be able to run that on one miseq run?

You might be able to tell them apart or you might not. It would probably take lots of computing and bioinformatics time, with no guarantee that you would get anything useful out at the end.

What kind of experiment are you doing, are you doing some kind of metagenomics experiment where there are lots of species all mixed together in one sample, or do you have separate samples from each individual?

How long are the genes that you want to look at in each sample, min, max?
And what is the total length of the different genes that you want to sequence?

[Sidsel]

Thanks again

I'm not doing metagenomics.

I want to look at 5 specific parasite genes, and their "flanking regions". This is because we look at microsatellite repeats and analyse the parasite lineage from that.

The genes vary in length from 2.1 kb to 5 kb, but including the flanking regions both upstream and downstream of the genes, we want to cover from 5 kb to 10 kb for each gene.

Does that amswer your question??

I wasn't aware that I could chose to add the barcodes after fragmentation! That helps alot - because then all the reads get barcoded. This would allow for sorting according to barcode, I suppose? And then subsequently cut the barcode and assemble the genes?

Just to clarify: I am aware that if I pool amplicons of the same gene from 100 samples, and label them all with the same barcode, then I won't be able to tell the different samples apart. But that's ok, because I think I'll still be able to perform a SNP analysis to see the degree of variation in the different genes?

Thanks again again!
Sidsel

[mastal]

I wasn't aware that I could chose to add the barcodes after fragmentation! That helps alot - because then all the reads get barcoded. This would allow for sorting according to barcode, I suppose? And then subsequently cut the barcode and assemble the genes?

The Illumina barcodes are read in separate reads from R1 and R2, and the reads should be sorted by barcode when you get them back from the sequencing center.

Just to clarify: I am aware that if I pool amplicons of the same gene from 100 samples, and label them all with the same barcode, then I won't be able to tell the different samples apart. But that's ok, because I think I'll still be able to perform a SNP analysis to see the degree of variation in the different genes?

I guess that's OK then, if you aren't concerned about having the particular sequence from each individual.

But if you are going to use a genomic DNA kit (Illumina TruSeq or Nextera) as opposed to a small amplicons kit, there's no point in running the different genes as different samples, you will have to to do either de novo assembly or alignment of your reads to a reference genome, and you can do that for all the genes at the same time.

Talk to your local Illumina rep, they will be able to advise you about the cost of different options.

[microgirl123]

I'm having a bit of trouble following what you want to do, but look into the Nextera XT kit from Illumina. This will fragment and tag your amplicons (each fragment of each individual sample will have the same dual index so you can sort them out later).