partial poor quality index and read2

[rogerzzw]

Hi, Guys

I am new to NGS. We've been trying to do whole genome bisulfite sequencing on Nextseq 500 for a few times with 2 samples.

We always encountered the same problem. After demultiplexing, there is about 50% of unknown index reads (most of them are "GGGGG" and "NNNNNN") . According to other centers using exact same kit and machine, they do not have any problems with adapters and indexes. And we always use single-use illumina kit and reagent for each run. So we do not know what happens.

We ve also tried to map read1 and read2 (with unknown barcodes). The alignment of read1 is quite good (90% aligniable), but reads2 is not good (~40%).

We used custom primer for read1 and illumina primers for read2 and index.

Many thanks in advance!

Roger

[pmiguel]

Not sure. Are you running just 2 samples per flowcell? Maybe you could drop down to one sample/flow cell and not use the index read?
How long are your reads?

--
Phillip

[GenoMax]

Phillip makes an excellent point. @rogerzzw: If your two indexes are very similar (can you post them) then you could essentially have a low nucleotide diversity/can't call bases type problem.

[pmiguel]

Ah, yeah. I was going to mention that possibility as well.
I've noticed recently that for the HiSeq, its ability to demultiplex single indexes seems to have plummeted with a couple of recent HCS upgrades. In principle, demultiplexing a single index is not of much importance. But it suggests that its ability to demultiplex two indexes that are not "well balanced" might also have dropped.

However, the issue you are having with read2 mapping at low rates makes it sound like an instrument issue.

Also, obviously, a NextSeq uses completely different chemistry than a HiSeq, so who knows?

Very funny that your failed indexes are either poly N or poly G as "G" in NextSeq chemistry is encoded as no fluorescence. I'm unclear how the instrument can distinguish between "N" -- no signal and "G" -- no signal. Although my Illumina sales rep assures me that these two states do not look the same...
--
Phillip

[rogerzzw]

Not sure. Are you running just 2 samples per flowcell? Maybe you could drop down to one sample/flow cell and not use the index read?
How long are your reads?

--
Phillip

Hi, Phillip.

I am always using 2 samples. I will use only one sample for next run to see what happens. We are doing 150bp PE

[GenoMax]

We had empirically determined that using 2 index pools (even though they may be well apart) was a recipe for problems. Three or more always seems to work best for HiSeq.

@Phillip: Interesting point about poly-G and poly-N.

[rogerzzw]

Phillip makes an excellent point. @rogerzzw: If your two indexes are very similar (can you post them) then you could essentially have a low nucleotide diversity/can't call bases type problem.

The barcodes I am using is
1. AGTGAG and GCACTA
2. AGTGAG and TGGTGA

But there are other groups suggesting they have no issue on barcodes, which makes me even more confused.

Thanks

[pmiguel]

The barcodes I am using is
1. AGTGAG and GCACTA
2. AGTGAG and TGGTGA

But there are other groups suggesting they have no issue on barcodes, which makes me even more confused.

Thanks

Right, we had no trouble with bar codes, even single ones, until the two most recent HCS updates. Now unbalanced ones seem to be problematic again.

How about the instrument that these are run on. Any issues with demultiplexing other projects?

--
Phillip

[tacana]

Hello all,

I was wondering what was the resolution of rogerzzw problem? We are having similar issues with our next seq. and would love to hear how was this resolved.

Thanks in advance,

Al